@alicloud/aidge20260428 5.7.0 → 5.8.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/client.d.ts +286 -78
- package/dist/client.js +405 -78
- package/dist/client.js.map +1 -1
- package/dist/models/AiTryOnRequest.d.ts +11 -2
- package/dist/models/AiTryOnRequest.js.map +1 -1
- package/dist/models/AiTryOnResponseBody.d.ts +35 -1
- package/dist/models/AiTryOnResponseBody.js.map +1 -1
- package/dist/models/DiduiAreaDeductionProRequest.d.ts +33 -0
- package/dist/models/DiduiAreaDeductionProRequest.js +60 -0
- package/dist/models/DiduiAreaDeductionProRequest.js.map +1 -0
- package/dist/models/DiduiAreaDeductionProResponse.d.ts +19 -0
- package/dist/models/DiduiAreaDeductionProResponse.js +69 -0
- package/dist/models/DiduiAreaDeductionProResponse.js.map +1 -0
- package/dist/models/DiduiAreaDeductionProResponseBody.d.ts +70 -0
- package/dist/models/DiduiAreaDeductionProResponseBody.js +88 -0
- package/dist/models/DiduiAreaDeductionProResponseBody.js.map +1 -0
- package/dist/models/GeneralRephotographyDetectionRequest.d.ts +23 -0
- package/dist/models/GeneralRephotographyDetectionRequest.js +58 -0
- package/dist/models/GeneralRephotographyDetectionRequest.js.map +1 -0
- package/dist/models/GeneralRephotographyDetectionResponse.d.ts +19 -0
- package/dist/models/GeneralRephotographyDetectionResponse.js +69 -0
- package/dist/models/GeneralRephotographyDetectionResponse.js.map +1 -0
- package/dist/models/GeneralRephotographyDetectionResponseBody.d.ts +170 -0
- package/dist/models/GeneralRephotographyDetectionResponseBody.js +172 -0
- package/dist/models/GeneralRephotographyDetectionResponseBody.js.map +1 -0
- package/dist/models/ImageTranslationPlusResponseBody.d.ts +1 -1
- package/dist/models/ImageTranslationProRequest.d.ts +3 -3
- package/dist/models/ImageTranslationStandardRequest.d.ts +10 -10
- package/dist/models/ImageTranslationStandardResponseBody.d.ts +8 -8
- package/dist/models/MerchandisePlacementDetectionProRequest.d.ts +39 -0
- package/dist/models/MerchandisePlacementDetectionProRequest.js +62 -0
- package/dist/models/MerchandisePlacementDetectionProRequest.js.map +1 -0
- package/dist/models/MerchandisePlacementDetectionProResponse.d.ts +19 -0
- package/dist/models/MerchandisePlacementDetectionProResponse.js +69 -0
- package/dist/models/MerchandisePlacementDetectionProResponse.js.map +1 -0
- package/dist/models/MerchandisePlacementDetectionProResponseBody.d.ts +118 -0
- package/dist/models/MerchandisePlacementDetectionProResponseBody.js +124 -0
- package/dist/models/MerchandisePlacementDetectionProResponseBody.js.map +1 -0
- package/dist/models/MerchandisePlacementDetectionRequest.d.ts +13 -5
- package/dist/models/MerchandisePlacementDetectionRequest.js +2 -0
- package/dist/models/MerchandisePlacementDetectionRequest.js.map +1 -1
- package/dist/models/MerchandisePlacementDetectionResponseBody.d.ts +7 -7
- package/dist/models/PetHealthAnalysisRequest.d.ts +36 -0
- package/dist/models/PetHealthAnalysisRequest.js +65 -0
- package/dist/models/PetHealthAnalysisRequest.js.map +1 -0
- package/dist/models/PetHealthAnalysisResponse.d.ts +19 -0
- package/dist/models/PetHealthAnalysisResponse.js +69 -0
- package/dist/models/PetHealthAnalysisResponse.js.map +1 -0
- package/dist/models/PetHealthAnalysisResponseBody.d.ts +70 -0
- package/dist/models/PetHealthAnalysisResponseBody.js +88 -0
- package/dist/models/PetHealthAnalysisResponseBody.js.map +1 -0
- package/dist/models/PetHealthAnalysisShrinkRequest.d.ts +36 -0
- package/dist/models/PetHealthAnalysisShrinkRequest.js +62 -0
- package/dist/models/PetHealthAnalysisShrinkRequest.js.map +1 -0
- package/dist/models/TextTranslateRequest.d.ts +6 -6
- package/dist/models/TextTranslateResponseBody.d.ts +1 -1
- package/dist/models/TextTranslateShrinkRequest.d.ts +6 -6
- package/dist/models/model.d.ts +21 -0
- package/dist/models/model.js +46 -4
- package/dist/models/model.js.map +1 -1
- package/package.json +1 -1
- package/src/client.ts +424 -78
- package/src/models/AiTryOnRequest.ts +11 -2
- package/src/models/AiTryOnResponseBody.ts +35 -1
- package/src/models/DiduiAreaDeductionProRequest.ts +48 -0
- package/src/models/DiduiAreaDeductionProResponse.ts +40 -0
- package/src/models/DiduiAreaDeductionProResponseBody.ts +104 -0
- package/src/models/GeneralRephotographyDetectionRequest.ts +36 -0
- package/src/models/GeneralRephotographyDetectionResponse.ts +40 -0
- package/src/models/GeneralRephotographyDetectionResponseBody.ts +259 -0
- package/src/models/ImageTranslationPlusResponseBody.ts +1 -1
- package/src/models/ImageTranslationProRequest.ts +3 -3
- package/src/models/ImageTranslationStandardRequest.ts +10 -10
- package/src/models/ImageTranslationStandardResponseBody.ts +8 -8
- package/src/models/MerchandisePlacementDetectionProRequest.ts +56 -0
- package/src/models/MerchandisePlacementDetectionProResponse.ts +40 -0
- package/src/models/MerchandisePlacementDetectionProResponseBody.ts +177 -0
- package/src/models/MerchandisePlacementDetectionRequest.ts +15 -5
- package/src/models/MerchandisePlacementDetectionResponseBody.ts +7 -7
- package/src/models/PetHealthAnalysisRequest.ts +56 -0
- package/src/models/PetHealthAnalysisResponse.ts +40 -0
- package/src/models/PetHealthAnalysisResponseBody.ts +104 -0
- package/src/models/PetHealthAnalysisShrinkRequest.ts +53 -0
- package/src/models/TextTranslateRequest.ts +6 -6
- package/src/models/TextTranslateResponseBody.ts +1 -1
- package/src/models/TextTranslateShrinkRequest.ts +6 -6
- package/src/models/model.ts +21 -0
package/dist/client.js
CHANGED
|
@@ -62,7 +62,38 @@ class Client extends openapi_core_1.default {
|
|
|
62
62
|
return openapi_core_2.OpenApiUtil.getEndpointRules(productId, regionId, endpointRule, network, suffix);
|
|
63
63
|
}
|
|
64
64
|
/**
|
|
65
|
-
*
|
|
65
|
+
* Virtual Try-On is an AI algorithm-based image generation API service that replaces the clothing on a model with the garment provided by the user based on the input garment image and model image, while preserving the model\\"s other details and scene details without altering the model\\"s pose, facial features, or other attributes, and achieving realistic and well-fitted clothing placement. Users only need to provide the URL of the original product image to obtain a processed clean image.
|
|
66
|
+
*
|
|
67
|
+
* @remarks
|
|
68
|
+
* ## Scenarios
|
|
69
|
+
* - **Virtual try-on for clothing products:**
|
|
70
|
+
* Supports multiple clothing categories including tops, bottoms, jumpsuits, and dresses. After uploading a person image and a product image, the service generates natural and realistic try-on results that intuitively showcase how the clothing looks when worn.
|
|
71
|
+
* - **Virtual try-on for shoes and hats:**
|
|
72
|
+
* Supports virtual try-on for shoes, hats, and similar products. The service intelligently matches the product position and proportion based on the person\\"s pose and body parts, producing coordinated and lifelike wearing effects.
|
|
73
|
+
* - **E-commerce product display and marketing:**
|
|
74
|
+
* Helps merchants quickly generate model try-on images for different products, reducing the costs of traditional photography, outfit changes, and post-production, and improving the production efficiency of product images.
|
|
75
|
+
* - **Online shopping experience optimization:**
|
|
76
|
+
* Consumers can preview how products look when worn through virtual try-on, gaining a more intuitive understanding of the product\\"s style, coordination, and overall appearance, which assists purchase decisions and enhances the shopping experience.
|
|
77
|
+
* ## Features
|
|
78
|
+
* - **First, the user needs to provide a model image and a product image. The system uses the model image as the try-on subject and applies the clothing from the product image onto the model:**
|
|
79
|
+
* - a. Model image: The person image on which the virtual try-on is performed.
|
|
80
|
+
* - b. Product image: The clothing, shoes, or hat image to be applied onto the model.
|
|
81
|
+
* - **Second, the user can select the corresponding product type based on the actual product. Supported types include tops, bottoms, jumpsuits, dresses, shoes, and hats:**
|
|
82
|
+
* - a. Specify product type: The user can directly specify the product type, and the system processes accordingly.
|
|
83
|
+
* - b. Automatic type recognition: If the user does not specify a product type, the system automatically identifies the product category from the product image and performs the virtual try-on based on the recognition result.
|
|
84
|
+
* - **Finally, the system combines the model image, product image, and product type to generate a virtual try-on image that showcases how the product looks on the model:**
|
|
85
|
+
* - a. Preserve model details: The generation process does not alter the person\\"s appearance, hairstyle, pose, body shape, or other details in the original model image.
|
|
86
|
+
* - b. Maintain scene consistency: The background, composition, lighting, and overall visual effect of the original model image remain unchanged. Only the wearing area corresponding to the product is processed.
|
|
87
|
+
* - c. Generate try-on results: The product is naturally applied onto the model, conforming to the person\\"s pose and body structure, producing a coordinated and realistic virtual try-on image.
|
|
88
|
+
* ## Sample results
|
|
89
|
+
* | **Type** | **Model image** | **Clothing image** | **Try-on result** |
|
|
90
|
+
* | --- | --- | --- | --- |
|
|
91
|
+
* | Tops |  |  |  |
|
|
92
|
+
* | Bottoms |  |  |  |
|
|
93
|
+
* | Shoes |  |  |  |
|
|
94
|
+
* | Jumpsuits |  |  |  |
|
|
95
|
+
* | Dresses |  |  |  |
|
|
96
|
+
* | Hats |  |  |  |
|
|
66
97
|
*
|
|
67
98
|
* @param request - AiTryOnRequest
|
|
68
99
|
* @param runtime - runtime options for this request RuntimeOptions
|
|
@@ -100,7 +131,38 @@ class Client extends openapi_core_1.default {
|
|
|
100
131
|
return $dara.cast(await this.callApi(params, req, runtime), new $_model.AiTryOnResponse({}));
|
|
101
132
|
}
|
|
102
133
|
/**
|
|
103
|
-
*
|
|
134
|
+
* Virtual Try-On is an AI algorithm-based image generation API service that replaces the clothing on a model with the garment provided by the user based on the input garment image and model image, while preserving the model\\"s other details and scene details without altering the model\\"s pose, facial features, or other attributes, and achieving realistic and well-fitted clothing placement. Users only need to provide the URL of the original product image to obtain a processed clean image.
|
|
135
|
+
*
|
|
136
|
+
* @remarks
|
|
137
|
+
* ## Scenarios
|
|
138
|
+
* - **Virtual try-on for clothing products:**
|
|
139
|
+
* Supports multiple clothing categories including tops, bottoms, jumpsuits, and dresses. After uploading a person image and a product image, the service generates natural and realistic try-on results that intuitively showcase how the clothing looks when worn.
|
|
140
|
+
* - **Virtual try-on for shoes and hats:**
|
|
141
|
+
* Supports virtual try-on for shoes, hats, and similar products. The service intelligently matches the product position and proportion based on the person\\"s pose and body parts, producing coordinated and lifelike wearing effects.
|
|
142
|
+
* - **E-commerce product display and marketing:**
|
|
143
|
+
* Helps merchants quickly generate model try-on images for different products, reducing the costs of traditional photography, outfit changes, and post-production, and improving the production efficiency of product images.
|
|
144
|
+
* - **Online shopping experience optimization:**
|
|
145
|
+
* Consumers can preview how products look when worn through virtual try-on, gaining a more intuitive understanding of the product\\"s style, coordination, and overall appearance, which assists purchase decisions and enhances the shopping experience.
|
|
146
|
+
* ## Features
|
|
147
|
+
* - **First, the user needs to provide a model image and a product image. The system uses the model image as the try-on subject and applies the clothing from the product image onto the model:**
|
|
148
|
+
* - a. Model image: The person image on which the virtual try-on is performed.
|
|
149
|
+
* - b. Product image: The clothing, shoes, or hat image to be applied onto the model.
|
|
150
|
+
* - **Second, the user can select the corresponding product type based on the actual product. Supported types include tops, bottoms, jumpsuits, dresses, shoes, and hats:**
|
|
151
|
+
* - a. Specify product type: The user can directly specify the product type, and the system processes accordingly.
|
|
152
|
+
* - b. Automatic type recognition: If the user does not specify a product type, the system automatically identifies the product category from the product image and performs the virtual try-on based on the recognition result.
|
|
153
|
+
* - **Finally, the system combines the model image, product image, and product type to generate a virtual try-on image that showcases how the product looks on the model:**
|
|
154
|
+
* - a. Preserve model details: The generation process does not alter the person\\"s appearance, hairstyle, pose, body shape, or other details in the original model image.
|
|
155
|
+
* - b. Maintain scene consistency: The background, composition, lighting, and overall visual effect of the original model image remain unchanged. Only the wearing area corresponding to the product is processed.
|
|
156
|
+
* - c. Generate try-on results: The product is naturally applied onto the model, conforming to the person\\"s pose and body structure, producing a coordinated and realistic virtual try-on image.
|
|
157
|
+
* ## Sample results
|
|
158
|
+
* | **Type** | **Model image** | **Clothing image** | **Try-on result** |
|
|
159
|
+
* | --- | --- | --- | --- |
|
|
160
|
+
* | Tops |  |  |  |
|
|
161
|
+
* | Bottoms |  |  |  |
|
|
162
|
+
* | Shoes |  |  |  |
|
|
163
|
+
* | Jumpsuits |  |  |  |
|
|
164
|
+
* | Dresses |  |  |  |
|
|
165
|
+
* | Hats |  |  |  |
|
|
104
166
|
*
|
|
105
167
|
* @param request - AiTryOnRequest
|
|
106
168
|
* @returns AiTryOnResponse
|
|
@@ -618,6 +680,70 @@ class Client extends openapi_core_1.default {
|
|
|
618
680
|
let runtime = new $dara.RuntimeOptions({});
|
|
619
681
|
return await this.diduiAreaDeductionWithOptions(request, runtime);
|
|
620
682
|
}
|
|
683
|
+
/**
|
|
684
|
+
* Analyzes the floor display area based on product recognition, SKU asset knowledge base, and a dedicated area reasoning model building. The product first locates floor display products, then reads the actual length, width, height, and reference images of the corresponding SKUs to infer the two side lengths and area of the bounding rectangle of the floor display footprint. This is applicable to floor display scale assessment and inspection review in retail stores.
|
|
685
|
+
*
|
|
686
|
+
* @remarks
|
|
687
|
+
* *Scenarios**
|
|
688
|
+
* - Floor display area assessment: Infers the bounding rectangle area of the floor display footprint based on an overall photo of the floor display and the actual dimensions of the products.
|
|
689
|
+
* - Multi-SKU mixed arrangement analysis: Uses the dimensions of each different product to participate in side length inference.
|
|
690
|
+
* - Area result review: Verifies results by combining the original image, product assets, and the returned two side lengths and area.
|
|
691
|
+
* **Features**
|
|
692
|
+
* - SKU asset-driven: Specifies the asset knowledge base through ragId. Each participating product must have a name, positive length, width, and height values, and at least one HTTPS reference image.
|
|
693
|
+
* - Phased area inference: Sequentially forms decision edges, SKU projections, dimension accumulation, and final area.
|
|
694
|
+
* - Unified units: SKU asset dimensions are stored in centimeters. Result side lengths are output in meters and area in square meters.
|
|
695
|
+
* - Completeness and numerical validation: Returns failure when there are missing assets, no valid bounding boxes, or the area does not satisfy product consistency.
|
|
696
|
+
*
|
|
697
|
+
* @param request - DiduiAreaDeductionProRequest
|
|
698
|
+
* @param runtime - runtime options for this request RuntimeOptions
|
|
699
|
+
* @returns DiduiAreaDeductionProResponse
|
|
700
|
+
*/
|
|
701
|
+
async diduiAreaDeductionProWithOptions(request, runtime) {
|
|
702
|
+
request.validate();
|
|
703
|
+
let query = {};
|
|
704
|
+
if (!$dara.isNull(request.imageUrl)) {
|
|
705
|
+
query["ImageUrl"] = request.imageUrl;
|
|
706
|
+
}
|
|
707
|
+
if (!$dara.isNull(request.ragId)) {
|
|
708
|
+
query["RagId"] = request.ragId;
|
|
709
|
+
}
|
|
710
|
+
let req = new openapi_core_2.$OpenApiUtil.OpenApiRequest({
|
|
711
|
+
query: openapi_core_2.OpenApiUtil.query(query),
|
|
712
|
+
});
|
|
713
|
+
let params = new openapi_core_2.$OpenApiUtil.Params({
|
|
714
|
+
action: "DiduiAreaDeductionPro",
|
|
715
|
+
version: "2026-04-28",
|
|
716
|
+
protocol: "HTTPS",
|
|
717
|
+
pathname: "/",
|
|
718
|
+
method: "POST",
|
|
719
|
+
authType: "AK",
|
|
720
|
+
style: "RPC",
|
|
721
|
+
reqBodyType: "formData",
|
|
722
|
+
bodyType: "json",
|
|
723
|
+
});
|
|
724
|
+
return $dara.cast(await this.callApi(params, req, runtime), new $_model.DiduiAreaDeductionProResponse({}));
|
|
725
|
+
}
|
|
726
|
+
/**
|
|
727
|
+
* Analyzes the floor display area based on product recognition, SKU asset knowledge base, and a dedicated area reasoning model building. The product first locates floor display products, then reads the actual length, width, height, and reference images of the corresponding SKUs to infer the two side lengths and area of the bounding rectangle of the floor display footprint. This is applicable to floor display scale assessment and inspection review in retail stores.
|
|
728
|
+
*
|
|
729
|
+
* @remarks
|
|
730
|
+
* *Scenarios**
|
|
731
|
+
* - Floor display area assessment: Infers the bounding rectangle area of the floor display footprint based on an overall photo of the floor display and the actual dimensions of the products.
|
|
732
|
+
* - Multi-SKU mixed arrangement analysis: Uses the dimensions of each different product to participate in side length inference.
|
|
733
|
+
* - Area result review: Verifies results by combining the original image, product assets, and the returned two side lengths and area.
|
|
734
|
+
* **Features**
|
|
735
|
+
* - SKU asset-driven: Specifies the asset knowledge base through ragId. Each participating product must have a name, positive length, width, and height values, and at least one HTTPS reference image.
|
|
736
|
+
* - Phased area inference: Sequentially forms decision edges, SKU projections, dimension accumulation, and final area.
|
|
737
|
+
* - Unified units: SKU asset dimensions are stored in centimeters. Result side lengths are output in meters and area in square meters.
|
|
738
|
+
* - Completeness and numerical validation: Returns failure when there are missing assets, no valid bounding boxes, or the area does not satisfy product consistency.
|
|
739
|
+
*
|
|
740
|
+
* @param request - DiduiAreaDeductionProRequest
|
|
741
|
+
* @returns DiduiAreaDeductionProResponse
|
|
742
|
+
*/
|
|
743
|
+
async diduiAreaDeductionPro(request) {
|
|
744
|
+
let runtime = new $dara.RuntimeOptions({});
|
|
745
|
+
return await this.diduiAreaDeductionProWithOptions(request, runtime);
|
|
746
|
+
}
|
|
621
747
|
/**
|
|
622
748
|
* Designed specifically for document translation. Supports translation between more than 400 language pairs (including bridged pairs) to handle multi-scenario, multi-page, and highly complex document translation. Scanned documents are not currently supported.
|
|
623
749
|
* Excels in the following areas:
|
|
@@ -798,6 +924,67 @@ class Client extends openapi_core_1.default {
|
|
|
798
924
|
let runtime = new $dara.RuntimeOptions({});
|
|
799
925
|
return await this.ecomVideoRecreationWithOptions(request, runtime);
|
|
800
926
|
}
|
|
927
|
+
/**
|
|
928
|
+
* Provides a general-purpose image review service built on recapture recognition and multimodal large model element localization. The service determines whether an image is a recaptured photo and locates primary physical products, screen display areas, printed photos, and visible recapture clues. It is applicable to risk screening and manual review of inspection photos across multiple store categories.
|
|
929
|
+
*
|
|
930
|
+
* @remarks
|
|
931
|
+
* *Scenarios**
|
|
932
|
+
* 1. Inspection photo review: Screens photos submitted by stores for recapture risks.
|
|
933
|
+
* 2. Screen or printed photo recapture verification: Locates visible screens, printed photos, and other carriers.
|
|
934
|
+
* 3. Review clue display: Annotates products and visible clues on the original image to assist re-examination.
|
|
935
|
+
* **Features**
|
|
936
|
+
* 1. General recapture detection: Returns an isFake Boolean result.
|
|
937
|
+
* 2. Review element localization: Distinguishes physical products, screens, printed photos, and visible recapture clues. Products displayed within a screen are not redundantly output as physical products.
|
|
938
|
+
* 3. Separate presentation of clues and conclusions: Localization text describes only visible phenomena. Normal reflections or standard packaging prints are not directly treated as recapture evidence.
|
|
939
|
+
* Supports a single HTTPS image as input. Supplementary localization returns up to 12 targets, each containing a category, a bounding box, and up to 24 characters of visible text. The coverage field indicates whether the localization is complete.
|
|
940
|
+
*
|
|
941
|
+
* @param request - GeneralRephotographyDetectionRequest
|
|
942
|
+
* @param runtime - runtime options for this request RuntimeOptions
|
|
943
|
+
* @returns GeneralRephotographyDetectionResponse
|
|
944
|
+
*/
|
|
945
|
+
async generalRephotographyDetectionWithOptions(request, runtime) {
|
|
946
|
+
request.validate();
|
|
947
|
+
let query = {};
|
|
948
|
+
if (!$dara.isNull(request.imageUrl)) {
|
|
949
|
+
query["ImageUrl"] = request.imageUrl;
|
|
950
|
+
}
|
|
951
|
+
let req = new openapi_core_2.$OpenApiUtil.OpenApiRequest({
|
|
952
|
+
query: openapi_core_2.OpenApiUtil.query(query),
|
|
953
|
+
});
|
|
954
|
+
let params = new openapi_core_2.$OpenApiUtil.Params({
|
|
955
|
+
action: "GeneralRephotographyDetection",
|
|
956
|
+
version: "2026-04-28",
|
|
957
|
+
protocol: "HTTPS",
|
|
958
|
+
pathname: "/",
|
|
959
|
+
method: "POST",
|
|
960
|
+
authType: "AK",
|
|
961
|
+
style: "RPC",
|
|
962
|
+
reqBodyType: "formData",
|
|
963
|
+
bodyType: "json",
|
|
964
|
+
});
|
|
965
|
+
return $dara.cast(await this.callApi(params, req, runtime), new $_model.GeneralRephotographyDetectionResponse({}));
|
|
966
|
+
}
|
|
967
|
+
/**
|
|
968
|
+
* Provides a general-purpose image review service built on recapture recognition and multimodal large model element localization. The service determines whether an image is a recaptured photo and locates primary physical products, screen display areas, printed photos, and visible recapture clues. It is applicable to risk screening and manual review of inspection photos across multiple store categories.
|
|
969
|
+
*
|
|
970
|
+
* @remarks
|
|
971
|
+
* *Scenarios**
|
|
972
|
+
* 1. Inspection photo review: Screens photos submitted by stores for recapture risks.
|
|
973
|
+
* 2. Screen or printed photo recapture verification: Locates visible screens, printed photos, and other carriers.
|
|
974
|
+
* 3. Review clue display: Annotates products and visible clues on the original image to assist re-examination.
|
|
975
|
+
* **Features**
|
|
976
|
+
* 1. General recapture detection: Returns an isFake Boolean result.
|
|
977
|
+
* 2. Review element localization: Distinguishes physical products, screens, printed photos, and visible recapture clues. Products displayed within a screen are not redundantly output as physical products.
|
|
978
|
+
* 3. Separate presentation of clues and conclusions: Localization text describes only visible phenomena. Normal reflections or standard packaging prints are not directly treated as recapture evidence.
|
|
979
|
+
* Supports a single HTTPS image as input. Supplementary localization returns up to 12 targets, each containing a category, a bounding box, and up to 24 characters of visible text. The coverage field indicates whether the localization is complete.
|
|
980
|
+
*
|
|
981
|
+
* @param request - GeneralRephotographyDetectionRequest
|
|
982
|
+
* @returns GeneralRephotographyDetectionResponse
|
|
983
|
+
*/
|
|
984
|
+
async generalRephotographyDetection(request) {
|
|
985
|
+
let runtime = new $dara.RuntimeOptions({});
|
|
986
|
+
return await this.generalRephotographyDetectionWithOptions(request, runtime);
|
|
987
|
+
}
|
|
801
988
|
/**
|
|
802
989
|
* Intelligently crops images to specified dimensions, automatically identifies the main subject area, and precisely crops to various sizes to meet design requirements across different scenarios. (Synchronous)
|
|
803
990
|
*
|
|
@@ -1514,13 +1701,13 @@ class Client extends openapi_core_1.default {
|
|
|
1514
1701
|
return await this.imageRemoveWithOptions(request, runtime);
|
|
1515
1702
|
}
|
|
1516
1703
|
/**
|
|
1517
|
-
*
|
|
1704
|
+
* Image Translation Plus is designed specifically for e-commerce images. It uses a Mixture of Experts (MOE) architecture and delivers superior translation accuracy for multiple minority languages compared to the Lite and Pro editions. We recommend using it for the following 12 language pairs, with more language pairs to be supported in the future.
|
|
1518
1705
|
*
|
|
1519
1706
|
* @remarks
|
|
1520
1707
|
* ## Product Introduction
|
|
1521
|
-
* Image Translation
|
|
1708
|
+
* Image Translation Plus is designed specifically for e-commerce images. It uses a Mixture of Experts (MOE) architecture and delivers superior translation accuracy for multiple minority languages compared to the Lite and Pro editions. We recommend using it for the following 8 language pairs, with more language pairs to be supported in the future.
|
|
1522
1709
|
* Supported language pairs:
|
|
1523
|
-
* | **
|
|
1710
|
+
* | **No.** | **Source language** | | **Target language** | |
|
|
1524
1711
|
* | --- | --- | --- | --- | --- |
|
|
1525
1712
|
* | | Language code | Language name | Language code | Language name |
|
|
1526
1713
|
* | 1 | en | English | ar | Arabic |
|
|
@@ -1542,7 +1729,7 @@ class Client extends openapi_core_1.default {
|
|
|
1542
1729
|
*
|
|
1543
1730
|
* * **Brand name protection**: Allows you to choose whether to translate brand names on images. This helps protect brand name information from being translated.
|
|
1544
1731
|
*
|
|
1545
|
-
* * **Translation intervention**: Allows you to customize translation results, including keeping text untranslated (ABC-ABC), specifying a translation (ABC-DEF), or removing text (ABC-empty value). This is commonly used for brand name protection and similar scenarios.
|
|
1732
|
+
* * **Translation intervention**: Allows you to customize translation results, including keeping text untranslated (ABC-ABC), specifying a translation (ABC-DEF), or removing text (ABC-empty value). This is commonly used for brand name protection and similar scenarios. Pass the corresponding intervention glossary ID when calling the API operation. This meets your translation needs across different scenarios. You can upload up to 100,000 intervention terms. If you need more, contact the platform for assistance.
|
|
1546
1733
|
*
|
|
1547
1734
|
* @param request - ImageTranslationPlusRequest
|
|
1548
1735
|
* @param runtime - runtime options for this request RuntimeOptions
|
|
@@ -1589,13 +1776,13 @@ class Client extends openapi_core_1.default {
|
|
|
1589
1776
|
return $dara.cast(await this.callApi(params, req, runtime), new $_model.ImageTranslationPlusResponse({}));
|
|
1590
1777
|
}
|
|
1591
1778
|
/**
|
|
1592
|
-
*
|
|
1779
|
+
* Image Translation Plus is designed specifically for e-commerce images. It uses a Mixture of Experts (MOE) architecture and delivers superior translation accuracy for multiple minority languages compared to the Lite and Pro editions. We recommend using it for the following 12 language pairs, with more language pairs to be supported in the future.
|
|
1593
1780
|
*
|
|
1594
1781
|
* @remarks
|
|
1595
1782
|
* ## Product Introduction
|
|
1596
|
-
* Image Translation
|
|
1783
|
+
* Image Translation Plus is designed specifically for e-commerce images. It uses a Mixture of Experts (MOE) architecture and delivers superior translation accuracy for multiple minority languages compared to the Lite and Pro editions. We recommend using it for the following 8 language pairs, with more language pairs to be supported in the future.
|
|
1597
1784
|
* Supported language pairs:
|
|
1598
|
-
* | **
|
|
1785
|
+
* | **No.** | **Source language** | | **Target language** | |
|
|
1599
1786
|
* | --- | --- | --- | --- | --- |
|
|
1600
1787
|
* | | Language code | Language name | Language code | Language name |
|
|
1601
1788
|
* | 1 | en | English | ar | Arabic |
|
|
@@ -1617,7 +1804,7 @@ class Client extends openapi_core_1.default {
|
|
|
1617
1804
|
*
|
|
1618
1805
|
* * **Brand name protection**: Allows you to choose whether to translate brand names on images. This helps protect brand name information from being translated.
|
|
1619
1806
|
*
|
|
1620
|
-
* * **Translation intervention**: Allows you to customize translation results, including keeping text untranslated (ABC-ABC), specifying a translation (ABC-DEF), or removing text (ABC-empty value). This is commonly used for brand name protection and similar scenarios.
|
|
1807
|
+
* * **Translation intervention**: Allows you to customize translation results, including keeping text untranslated (ABC-ABC), specifying a translation (ABC-DEF), or removing text (ABC-empty value). This is commonly used for brand name protection and similar scenarios. Pass the corresponding intervention glossary ID when calling the API operation. This meets your translation needs across different scenarios. You can upload up to 100,000 intervention terms. If you need more, contact the platform for assistance.
|
|
1621
1808
|
*
|
|
1622
1809
|
* @param request - ImageTranslationPlusRequest
|
|
1623
1810
|
* @returns ImageTranslationPlusResponse
|
|
@@ -1627,18 +1814,18 @@ class Client extends openapi_core_1.default {
|
|
|
1627
1814
|
return await this.imageTranslationPlusWithOptions(request, runtime);
|
|
1628
1815
|
}
|
|
1629
1816
|
/**
|
|
1630
|
-
* Image Translation Pro is designed for e-commerce images. It integrates multimodal
|
|
1817
|
+
* Image Translation Pro is designed for e-commerce images. It integrates multimodal foundation model technology to achieve more accurate image understanding, significantly improve translation quality, and continuously expand and optimize multilingual translation capabilities. It supports more than 400 language directions (including bridged directions).
|
|
1631
1818
|
*
|
|
1632
1819
|
* @remarks
|
|
1633
1820
|
* ## Product Introduction
|
|
1634
|
-
* Image Translation Pro is designed for e-commerce images. It integrates multimodal
|
|
1821
|
+
* Image Translation Pro is designed for e-commerce images. It integrates multimodal foundation model technology to achieve more accurate image understanding, significantly improve translation quality, and continuously expand and optimize multilingual translation capabilities. It supports more than 400 language directions (including bridged directions).
|
|
1635
1822
|
* ## Scenarios
|
|
1636
1823
|
* E-commerce product images, marketing images, and images for various other scenarios.
|
|
1637
1824
|
* ## Features
|
|
1638
1825
|
* - **Product subject protection**: Specify whether to translate text on the product subject. This helps you protect subject information from being translated, such as embedded information like product names.
|
|
1639
|
-
* - **Post-translation editing**: Specify whether to return layout information such as text position, font, and color. This can be used for secondary editing when integrated with an image editor. The editor SDK is not yet available. Follow platform notifications for updates.
|
|
1826
|
+
* - **Post-translation editing**: Specify whether to return layout information such as text position, font, and color. This can be used for secondary editing when integrated with an image editor. The editor SDK package is not yet available. Follow platform notifications for updates.
|
|
1640
1827
|
* - **Brand name protection**: Specify whether to translate brand names on images. This helps you protect brand name information from being translated.
|
|
1641
|
-
* - **Translation intervention**: Customize translation results, including do-not-translate (ABC→ABC), specified translation (ABC→DEF), and remove text (ABC→empty value). This is commonly used for brand name protection and similar scenarios. Pass the corresponding intervention glossary ID when calling the API
|
|
1828
|
+
* - **Translation intervention**: Customize translation results, including do-not-translate (ABC→ABC), specified translation (ABC→DEF), and remove text (ABC→empty value). This is commonly used for brand name protection and similar scenarios. Pass the corresponding intervention glossary ID when calling the API. You can upload up to 100,000 intervention terms. If you need more, contact the platform for assistance.
|
|
1642
1829
|
* - <props="china"> **Image editor integration**: For more information, see [Image Translation Editor Protocol](https://www.alibabacloud.com/help/en/document_detail/3042647.html).
|
|
1643
1830
|
*
|
|
1644
1831
|
* @param request - ImageTranslationProRequest
|
|
@@ -1689,18 +1876,18 @@ class Client extends openapi_core_1.default {
|
|
|
1689
1876
|
return $dara.cast(await this.callApi(params, req, runtime), new $_model.ImageTranslationProResponse({}));
|
|
1690
1877
|
}
|
|
1691
1878
|
/**
|
|
1692
|
-
* Image Translation Pro is designed for e-commerce images. It integrates multimodal
|
|
1879
|
+
* Image Translation Pro is designed for e-commerce images. It integrates multimodal foundation model technology to achieve more accurate image understanding, significantly improve translation quality, and continuously expand and optimize multilingual translation capabilities. It supports more than 400 language directions (including bridged directions).
|
|
1693
1880
|
*
|
|
1694
1881
|
* @remarks
|
|
1695
1882
|
* ## Product Introduction
|
|
1696
|
-
* Image Translation Pro is designed for e-commerce images. It integrates multimodal
|
|
1883
|
+
* Image Translation Pro is designed for e-commerce images. It integrates multimodal foundation model technology to achieve more accurate image understanding, significantly improve translation quality, and continuously expand and optimize multilingual translation capabilities. It supports more than 400 language directions (including bridged directions).
|
|
1697
1884
|
* ## Scenarios
|
|
1698
1885
|
* E-commerce product images, marketing images, and images for various other scenarios.
|
|
1699
1886
|
* ## Features
|
|
1700
1887
|
* - **Product subject protection**: Specify whether to translate text on the product subject. This helps you protect subject information from being translated, such as embedded information like product names.
|
|
1701
|
-
* - **Post-translation editing**: Specify whether to return layout information such as text position, font, and color. This can be used for secondary editing when integrated with an image editor. The editor SDK is not yet available. Follow platform notifications for updates.
|
|
1888
|
+
* - **Post-translation editing**: Specify whether to return layout information such as text position, font, and color. This can be used for secondary editing when integrated with an image editor. The editor SDK package is not yet available. Follow platform notifications for updates.
|
|
1702
1889
|
* - **Brand name protection**: Specify whether to translate brand names on images. This helps you protect brand name information from being translated.
|
|
1703
|
-
* - **Translation intervention**: Customize translation results, including do-not-translate (ABC→ABC), specified translation (ABC→DEF), and remove text (ABC→empty value). This is commonly used for brand name protection and similar scenarios. Pass the corresponding intervention glossary ID when calling the API
|
|
1890
|
+
* - **Translation intervention**: Customize translation results, including do-not-translate (ABC→ABC), specified translation (ABC→DEF), and remove text (ABC→empty value). This is commonly used for brand name protection and similar scenarios. Pass the corresponding intervention glossary ID when calling the API. You can upload up to 100,000 intervention terms. If you need more, contact the platform for assistance.
|
|
1704
1891
|
* - <props="china"> **Image editor integration**: For more information, see [Image Translation Editor Protocol](https://www.alibabacloud.com/help/en/document_detail/3042647.html).
|
|
1705
1892
|
*
|
|
1706
1893
|
* @param request - ImageTranslationProRequest
|
|
@@ -1711,21 +1898,21 @@ class Client extends openapi_core_1.default {
|
|
|
1711
1898
|
return await this.imageTranslationProWithOptions(request, runtime);
|
|
1712
1899
|
}
|
|
1713
1900
|
/**
|
|
1714
|
-
* Image Translation is designed specifically for e-commerce images. It supports more than
|
|
1901
|
+
* Image Translation is designed specifically for e-commerce images. It supports more than 400 language directions (including bridging), accurately processes complex text layouts within images, and ensures that translated content matches the image content. This helps e-commerce platforms and developers easily achieve multilingual conversion and rendering of image content. The product also returns layer information such as text position, font, and color within images, which facilitates secondary editing of translation results through an image editor.
|
|
1715
1902
|
*
|
|
1716
1903
|
* @remarks
|
|
1717
1904
|
* # 1. Product Introduction
|
|
1718
|
-
* Image Translation is designed specifically for e-commerce images. It supports more than 100 language directions (including bridging), accurately processes complex text layouts within images, and ensures that translated content matches the image content. This helps e-commerce platforms and developers easily achieve multilingual conversion and rendering of image content. The product also returns layer information such as text position, font, and color within images,
|
|
1719
|
-
* # 2.
|
|
1905
|
+
* Image Translation is designed specifically for e-commerce images. It supports more than 100 language directions (including bridging), accurately processes complex text layouts within images, and ensures that translated content matches the image content. This helps e-commerce platforms and developers easily achieve multilingual conversion and rendering of image content. The product also returns layer information such as text position, font, and color within images, which facilitates secondary editing of translation results through an image editor.
|
|
1906
|
+
* # 2. Common scenarios
|
|
1720
1907
|
* E-commerce product images, marketing images, and images for various other scenarios.
|
|
1721
1908
|
* # 3. Features
|
|
1722
|
-
* * **Product area protection**:
|
|
1909
|
+
* * **Product area protection**: You can choose whether to translate text on the product area of an image. This helps protect product area information from being translated, such as embedded product names.
|
|
1723
1910
|
*
|
|
1724
|
-
* * **Brand name protection**:
|
|
1911
|
+
* * **Brand name protection**: You can choose whether to translate brand names on images. This helps protect brand name information from being translated.
|
|
1725
1912
|
*
|
|
1726
|
-
* * **Secondary editing after translation (layer information)**:
|
|
1913
|
+
* * **Secondary editing after translation (layer information)**: You can choose whether to return layer information such as text position, font, and color. This information can be used to connect to an image editor for secondary editing of translation results. For the release progress of the editor SDK package, follow platform notifications.
|
|
1727
1914
|
*
|
|
1728
|
-
* * **Translation intervention**:
|
|
1915
|
+
* * **Translation intervention**: You can customize translation results, including keeping text untranslated (ABC-ABC), specifying translations (ABC-DEF), or removing text (ABC-empty value). This is commonly used for scenarios such as brand name protection. Simply pass the corresponding intervention glossary ID when calling the API operation. This meets your translation needs across different scenarios. You can upload up to 100,000 intervention terms. If you need more, contact the platform for assistance.
|
|
1729
1916
|
* * **Image editor integration**: For more information, see [Image Translation editor protocol](https://www.alibabacloud.com/help/en/document_detail/3042647.html).
|
|
1730
1917
|
*
|
|
1731
1918
|
* @param request - ImageTranslationStandardRequest
|
|
@@ -1773,21 +1960,21 @@ class Client extends openapi_core_1.default {
|
|
|
1773
1960
|
return $dara.cast(await this.callApi(params, req, runtime), new $_model.ImageTranslationStandardResponse({}));
|
|
1774
1961
|
}
|
|
1775
1962
|
/**
|
|
1776
|
-
* Image Translation is designed specifically for e-commerce images. It supports more than
|
|
1963
|
+
* Image Translation is designed specifically for e-commerce images. It supports more than 400 language directions (including bridging), accurately processes complex text layouts within images, and ensures that translated content matches the image content. This helps e-commerce platforms and developers easily achieve multilingual conversion and rendering of image content. The product also returns layer information such as text position, font, and color within images, which facilitates secondary editing of translation results through an image editor.
|
|
1777
1964
|
*
|
|
1778
1965
|
* @remarks
|
|
1779
1966
|
* # 1. Product Introduction
|
|
1780
|
-
* Image Translation is designed specifically for e-commerce images. It supports more than 100 language directions (including bridging), accurately processes complex text layouts within images, and ensures that translated content matches the image content. This helps e-commerce platforms and developers easily achieve multilingual conversion and rendering of image content. The product also returns layer information such as text position, font, and color within images,
|
|
1781
|
-
* # 2.
|
|
1967
|
+
* Image Translation is designed specifically for e-commerce images. It supports more than 100 language directions (including bridging), accurately processes complex text layouts within images, and ensures that translated content matches the image content. This helps e-commerce platforms and developers easily achieve multilingual conversion and rendering of image content. The product also returns layer information such as text position, font, and color within images, which facilitates secondary editing of translation results through an image editor.
|
|
1968
|
+
* # 2. Common scenarios
|
|
1782
1969
|
* E-commerce product images, marketing images, and images for various other scenarios.
|
|
1783
1970
|
* # 3. Features
|
|
1784
|
-
* * **Product area protection**:
|
|
1971
|
+
* * **Product area protection**: You can choose whether to translate text on the product area of an image. This helps protect product area information from being translated, such as embedded product names.
|
|
1785
1972
|
*
|
|
1786
|
-
* * **Brand name protection**:
|
|
1973
|
+
* * **Brand name protection**: You can choose whether to translate brand names on images. This helps protect brand name information from being translated.
|
|
1787
1974
|
*
|
|
1788
|
-
* * **Secondary editing after translation (layer information)**:
|
|
1975
|
+
* * **Secondary editing after translation (layer information)**: You can choose whether to return layer information such as text position, font, and color. This information can be used to connect to an image editor for secondary editing of translation results. For the release progress of the editor SDK package, follow platform notifications.
|
|
1789
1976
|
*
|
|
1790
|
-
* * **Translation intervention**:
|
|
1977
|
+
* * **Translation intervention**: You can customize translation results, including keeping text untranslated (ABC-ABC), specifying translations (ABC-DEF), or removing text (ABC-empty value). This is commonly used for scenarios such as brand name protection. Simply pass the corresponding intervention glossary ID when calling the API operation. This meets your translation needs across different scenarios. You can upload up to 100,000 intervention terms. If you need more, contact the platform for assistance.
|
|
1791
1978
|
* * **Image editor integration**: For more information, see [Image Translation editor protocol](https://www.alibabacloud.com/help/en/document_detail/3042647.html).
|
|
1792
1979
|
*
|
|
1793
1980
|
* @param request - ImageTranslationStandardRequest
|
|
@@ -1943,25 +2130,25 @@ class Client extends openapi_core_1.default {
|
|
|
1943
2130
|
return await this.materialInspectionWithOptions(request, runtime);
|
|
1944
2131
|
}
|
|
1945
2132
|
/**
|
|
1946
|
-
* Detects and identifies products displayed on store shelves or floor stacks by
|
|
2133
|
+
* Detects and identifies products displayed on store shelves or floor stacks by using a multimodal large language model (MLLM) combined with visual vector recall. The service is designed for shelf and floor stack inspection scenarios in offline retail and channel stores. It first uses the MLLM to detect the bounding box and preliminary name of each product in a store photo, then performs similarity recall against a customer-specific SKU vector library to exactly match each product to the customer\\"s own SKUs, and outputs per-product structured results that can be directly integrated with business systems. This service is suitable for large-scale, high-concurrent visual recognition pipelines such as remote display auditing by headquarters, SKU distribution verification, and display compliance inspection for offline stores.
|
|
1947
2134
|
*
|
|
1948
2135
|
* @remarks
|
|
1949
2136
|
* ## **Product Introduction**
|
|
1950
|
-
* Display detection is an intelligent
|
|
2137
|
+
* Display detection is an intelligent product display recognition capability built on a multimodal large language model (MLLM) combined with visual vector recall. It is designed for shelf and floor stack inspection scenarios in offline retail and channel stores. The service takes real photos of store shelves and floor stacks, first uses the MLLM to detect the bounding box and preliminary name of each product, then performs similarity recall against a customer-specific SKU vector library to exactly match each product to the customer\\"s own SKUs, and outputs per-product structured results that can be directly integrated with business systems. This service is suitable for large-scale, high-concurrent visual recognition pipelines such as remote display auditing by headquarters, SKU distribution verification, and display compliance inspection for offline stores.
|
|
1951
2138
|
* ## **Scenarios**
|
|
1952
|
-
* - **Shelf/floor
|
|
2139
|
+
* - **Shelf/floor stack display recognition and inventory**: Identifies each product in a shelf or floor stack image, matches it to the customer\\"s SKU library, and outputs a per-product list (product name + position + similarity score), replacing manual counting.
|
|
1953
2140
|
*
|
|
1954
|
-
* - **Distribution verification/SKU compliance**: Verifies whether a store displays the specified SKUs as required by headquarters and whether any out-of-stock or misplaced items exist
|
|
2141
|
+
* - **Distribution verification/SKU compliance**: Verifies whether a store displays the specified SKUs as required by headquarters and whether any out-of-stock or misplaced items exist. This supports batch automated verification during sales promotions or new product launches.
|
|
1955
2142
|
*
|
|
1956
|
-
* - **Customer-specific SKU library**: Customers can upload their own standard SKU images to build a dedicated vector library (isolated by `RagId`). During recognition,
|
|
1957
|
-
* ## **
|
|
1958
|
-
* - **Two-stage detection +
|
|
2143
|
+
* - **Customer-specific SKU library**: Customers can upload their own standard SKU images to build a dedicated vector library (isolated by `RagId`). During recognition, retrieval is performed only within the corresponding customer library, ensuring name accuracy and data isolation between customers.
|
|
2144
|
+
* ## **Functions and features**
|
|
2145
|
+
* - **Two-stage detection + retrieval pipeline**: The MLLM detects each product bounding box and preliminary name across the entire image (semantic-level understanding, adapted for dense displays and partial occlusion). Then, visual embedding vectors are used for similarity retrieval within the customer\\"s SKU library, and the top-1 precise SKU name overrides the preliminary name, balancing recall rate and name accuracy.
|
|
1959
2146
|
*
|
|
1960
|
-
* - **
|
|
2147
|
+
* - **Isolated retrieval with dedicated vector libraries**: A dedicated SKU vector library is maintained for each customer by `RagId`. Recognition retrieves only within the corresponding library, ensuring strict isolation between customers. Newly added or updated SKUs take effect in real time.
|
|
1961
2148
|
*
|
|
1962
|
-
* - **Per-product structured output**: Output is unified as `Result.Items[]`, with each product independently providing a bounding box, recalled SKU name, SKU code, and similarity score. This facilitates direct integration with downstream inventory
|
|
2149
|
+
* - **Per-product structured output**: Output is unified as `Result.Items[]`, with each product independently providing a bounding box, recalled SKU name, SKU code, and similarity score. This facilitates direct integration with downstream inventory and verification systems and supports case-level manual review and traceability.
|
|
1963
2150
|
*
|
|
1964
|
-
* - **Supported input formats**: Currently supports image URLs accessible over the Internet or
|
|
2151
|
+
* - **Supported input formats**: Currently supports image URLs accessible over the Internet or through OSS. A single shelf or floor stack image is sufficient for recognition, and the output structure is consistent.
|
|
1965
2152
|
*
|
|
1966
2153
|
* @param request - MerchandisePlacementDetectionRequest
|
|
1967
2154
|
* @param runtime - runtime options for this request RuntimeOptions
|
|
@@ -1979,6 +2166,9 @@ class Client extends openapi_core_1.default {
|
|
|
1979
2166
|
if (!$dara.isNull(request.ragId)) {
|
|
1980
2167
|
query["RagId"] = request.ragId;
|
|
1981
2168
|
}
|
|
2169
|
+
if (!$dara.isNull(request.rule)) {
|
|
2170
|
+
query["Rule"] = request.rule;
|
|
2171
|
+
}
|
|
1982
2172
|
if (!$dara.isNull(request.type)) {
|
|
1983
2173
|
query["Type"] = request.type;
|
|
1984
2174
|
}
|
|
@@ -1999,25 +2189,25 @@ class Client extends openapi_core_1.default {
|
|
|
1999
2189
|
return $dara.cast(await this.callApi(params, req, runtime), new $_model.MerchandisePlacementDetectionResponse({}));
|
|
2000
2190
|
}
|
|
2001
2191
|
/**
|
|
2002
|
-
* Detects and identifies products displayed on store shelves or floor stacks by
|
|
2192
|
+
* Detects and identifies products displayed on store shelves or floor stacks by using a multimodal large language model (MLLM) combined with visual vector recall. The service is designed for shelf and floor stack inspection scenarios in offline retail and channel stores. It first uses the MLLM to detect the bounding box and preliminary name of each product in a store photo, then performs similarity recall against a customer-specific SKU vector library to exactly match each product to the customer\\"s own SKUs, and outputs per-product structured results that can be directly integrated with business systems. This service is suitable for large-scale, high-concurrent visual recognition pipelines such as remote display auditing by headquarters, SKU distribution verification, and display compliance inspection for offline stores.
|
|
2003
2193
|
*
|
|
2004
2194
|
* @remarks
|
|
2005
2195
|
* ## **Product Introduction**
|
|
2006
|
-
* Display detection is an intelligent
|
|
2196
|
+
* Display detection is an intelligent product display recognition capability built on a multimodal large language model (MLLM) combined with visual vector recall. It is designed for shelf and floor stack inspection scenarios in offline retail and channel stores. The service takes real photos of store shelves and floor stacks, first uses the MLLM to detect the bounding box and preliminary name of each product, then performs similarity recall against a customer-specific SKU vector library to exactly match each product to the customer\\"s own SKUs, and outputs per-product structured results that can be directly integrated with business systems. This service is suitable for large-scale, high-concurrent visual recognition pipelines such as remote display auditing by headquarters, SKU distribution verification, and display compliance inspection for offline stores.
|
|
2007
2197
|
* ## **Scenarios**
|
|
2008
|
-
* - **Shelf/floor
|
|
2198
|
+
* - **Shelf/floor stack display recognition and inventory**: Identifies each product in a shelf or floor stack image, matches it to the customer\\"s SKU library, and outputs a per-product list (product name + position + similarity score), replacing manual counting.
|
|
2009
2199
|
*
|
|
2010
|
-
* - **Distribution verification/SKU compliance**: Verifies whether a store displays the specified SKUs as required by headquarters and whether any out-of-stock or misplaced items exist
|
|
2200
|
+
* - **Distribution verification/SKU compliance**: Verifies whether a store displays the specified SKUs as required by headquarters and whether any out-of-stock or misplaced items exist. This supports batch automated verification during sales promotions or new product launches.
|
|
2011
2201
|
*
|
|
2012
|
-
* - **Customer-specific SKU library**: Customers can upload their own standard SKU images to build a dedicated vector library (isolated by `RagId`). During recognition,
|
|
2013
|
-
* ## **
|
|
2014
|
-
* - **Two-stage detection +
|
|
2202
|
+
* - **Customer-specific SKU library**: Customers can upload their own standard SKU images to build a dedicated vector library (isolated by `RagId`). During recognition, retrieval is performed only within the corresponding customer library, ensuring name accuracy and data isolation between customers.
|
|
2203
|
+
* ## **Functions and features**
|
|
2204
|
+
* - **Two-stage detection + retrieval pipeline**: The MLLM detects each product bounding box and preliminary name across the entire image (semantic-level understanding, adapted for dense displays and partial occlusion). Then, visual embedding vectors are used for similarity retrieval within the customer\\"s SKU library, and the top-1 precise SKU name overrides the preliminary name, balancing recall rate and name accuracy.
|
|
2015
2205
|
*
|
|
2016
|
-
* - **
|
|
2206
|
+
* - **Isolated retrieval with dedicated vector libraries**: A dedicated SKU vector library is maintained for each customer by `RagId`. Recognition retrieves only within the corresponding library, ensuring strict isolation between customers. Newly added or updated SKUs take effect in real time.
|
|
2017
2207
|
*
|
|
2018
|
-
* - **Per-product structured output**: Output is unified as `Result.Items[]`, with each product independently providing a bounding box, recalled SKU name, SKU code, and similarity score. This facilitates direct integration with downstream inventory
|
|
2208
|
+
* - **Per-product structured output**: Output is unified as `Result.Items[]`, with each product independently providing a bounding box, recalled SKU name, SKU code, and similarity score. This facilitates direct integration with downstream inventory and verification systems and supports case-level manual review and traceability.
|
|
2019
2209
|
*
|
|
2020
|
-
* - **Supported input formats**: Currently supports image URLs accessible over the Internet or
|
|
2210
|
+
* - **Supported input formats**: Currently supports image URLs accessible over the Internet or through OSS. A single shelf or floor stack image is sufficient for recognition, and the output structure is consistent.
|
|
2021
2211
|
*
|
|
2022
2212
|
* @param request - MerchandisePlacementDetectionRequest
|
|
2023
2213
|
* @returns MerchandisePlacementDetectionResponse
|
|
@@ -2026,6 +2216,73 @@ class Client extends openapi_core_1.default {
|
|
|
2026
2216
|
let runtime = new $dara.RuntimeOptions({});
|
|
2027
2217
|
return await this.merchandisePlacementDetectionWithOptions(request, runtime);
|
|
2028
2218
|
}
|
|
2219
|
+
/**
|
|
2220
|
+
* Detects the positions and names of visible products in an image based on a dedicated visual model. The product supports default recognition rules for the Genki Forest business type and custom rules provided by the invoker. It outputs per-product positions and names, applicable to beverage display inspection and product location display.
|
|
2221
|
+
*
|
|
2222
|
+
* @remarks
|
|
2223
|
+
* ## Scenarios
|
|
2224
|
+
* - Beverage display recognition: Uses default rules for the Genki Forest business type to recognize visible beverage products.
|
|
2225
|
+
* - Custom detection rules: Specifies product recognition tasks that comply with output constraints through the rule parameter.
|
|
2226
|
+
* - Product location display: Uses per-product names and bounding boxes for display, statistics, or manual review.
|
|
2227
|
+
* ## Features
|
|
2228
|
+
* - Direct product recognition: Each product outputs a bounding box and a name. The name is returned as unknown if it cannot be determined.
|
|
2229
|
+
* - Two rule entry points: A non-empty rule takes priority. If no valid rule is provided, the type must be set to Genki Forest.
|
|
2230
|
+
* - Unified result validation: Bounding box coordinates are converted to integers and restricted to the range 0–1000. Reversed coordinates are swapped. Missing names default to unknown. The entire batch fails if any item is not a valid product object, has an invalid bounding box, or has no area.
|
|
2231
|
+
* - Result scope: Currently outputs detected names and positions only. Vector library recall, SKU codes, and similarity scores are not included.
|
|
2232
|
+
*
|
|
2233
|
+
* @param request - MerchandisePlacementDetectionProRequest
|
|
2234
|
+
* @param runtime - runtime options for this request RuntimeOptions
|
|
2235
|
+
* @returns MerchandisePlacementDetectionProResponse
|
|
2236
|
+
*/
|
|
2237
|
+
async merchandisePlacementDetectionProWithOptions(request, runtime) {
|
|
2238
|
+
request.validate();
|
|
2239
|
+
let query = {};
|
|
2240
|
+
if (!$dara.isNull(request.imageUrl)) {
|
|
2241
|
+
query["ImageUrl"] = request.imageUrl;
|
|
2242
|
+
}
|
|
2243
|
+
if (!$dara.isNull(request.rule)) {
|
|
2244
|
+
query["Rule"] = request.rule;
|
|
2245
|
+
}
|
|
2246
|
+
if (!$dara.isNull(request.type)) {
|
|
2247
|
+
query["Type"] = request.type;
|
|
2248
|
+
}
|
|
2249
|
+
let req = new openapi_core_2.$OpenApiUtil.OpenApiRequest({
|
|
2250
|
+
query: openapi_core_2.OpenApiUtil.query(query),
|
|
2251
|
+
});
|
|
2252
|
+
let params = new openapi_core_2.$OpenApiUtil.Params({
|
|
2253
|
+
action: "MerchandisePlacementDetectionPro",
|
|
2254
|
+
version: "2026-04-28",
|
|
2255
|
+
protocol: "HTTPS",
|
|
2256
|
+
pathname: "/",
|
|
2257
|
+
method: "POST",
|
|
2258
|
+
authType: "AK",
|
|
2259
|
+
style: "RPC",
|
|
2260
|
+
reqBodyType: "formData",
|
|
2261
|
+
bodyType: "json",
|
|
2262
|
+
});
|
|
2263
|
+
return $dara.cast(await this.callApi(params, req, runtime), new $_model.MerchandisePlacementDetectionProResponse({}));
|
|
2264
|
+
}
|
|
2265
|
+
/**
|
|
2266
|
+
* Detects the positions and names of visible products in an image based on a dedicated visual model. The product supports default recognition rules for the Genki Forest business type and custom rules provided by the invoker. It outputs per-product positions and names, applicable to beverage display inspection and product location display.
|
|
2267
|
+
*
|
|
2268
|
+
* @remarks
|
|
2269
|
+
* ## Scenarios
|
|
2270
|
+
* - Beverage display recognition: Uses default rules for the Genki Forest business type to recognize visible beverage products.
|
|
2271
|
+
* - Custom detection rules: Specifies product recognition tasks that comply with output constraints through the rule parameter.
|
|
2272
|
+
* - Product location display: Uses per-product names and bounding boxes for display, statistics, or manual review.
|
|
2273
|
+
* ## Features
|
|
2274
|
+
* - Direct product recognition: Each product outputs a bounding box and a name. The name is returned as unknown if it cannot be determined.
|
|
2275
|
+
* - Two rule entry points: A non-empty rule takes priority. If no valid rule is provided, the type must be set to Genki Forest.
|
|
2276
|
+
* - Unified result validation: Bounding box coordinates are converted to integers and restricted to the range 0–1000. Reversed coordinates are swapped. Missing names default to unknown. The entire batch fails if any item is not a valid product object, has an invalid bounding box, or has no area.
|
|
2277
|
+
* - Result scope: Currently outputs detected names and positions only. Vector library recall, SKU codes, and similarity scores are not included.
|
|
2278
|
+
*
|
|
2279
|
+
* @param request - MerchandisePlacementDetectionProRequest
|
|
2280
|
+
* @returns MerchandisePlacementDetectionProResponse
|
|
2281
|
+
*/
|
|
2282
|
+
async merchandisePlacementDetectionPro(request) {
|
|
2283
|
+
let runtime = new $dara.RuntimeOptions({});
|
|
2284
|
+
return await this.merchandisePlacementDetectionProWithOptions(request, runtime);
|
|
2285
|
+
}
|
|
2029
2286
|
/**
|
|
2030
2287
|
* An intelligent logistics parcel review product built on the collaboration of a multimodal large language model (MLLM) and dedicated vision algorithms. It targets automated compliance review for the "piece-weight-dimension" stage in e-commerce logistics. The product performs structured review across multiple dimensions, including bounding box validity, bounding box fit, and foreign object interference for parcels on scanning platforms. It supports different hardware capture scenarios such as white and green background boards, and outputs Boolean review conclusions with Chinese-language reason descriptions that can be directly integrated with business systems. It is suitable for large-scale, high-concurrency inbound quality inspection pipelines.
|
|
2031
2288
|
*
|
|
@@ -2098,6 +2355,76 @@ class Client extends openapi_core_1.default {
|
|
|
2098
2355
|
let runtime = new $dara.RuntimeOptions({});
|
|
2099
2356
|
return await this.packageWeightSizeCheckWithOptions(request, runtime);
|
|
2100
2357
|
}
|
|
2358
|
+
/**
|
|
2359
|
+
* Analyzes pet excrement images using a pet multimodal model building product. Supports uploading pet excrement images, determines by default whether excrement is present in the image, locates and performs categorization on each visible excrement area, and returns model-generated text containing the categorization results. The invoker can also optionally provide a user prompt and a system prompt to customize image analysis requirements and response format.
|
|
2360
|
+
* Applicable scenarios:
|
|
2361
|
+
* 1. Excrement image analysis: Upload pet excrement images for analysis using the default method.
|
|
2362
|
+
* 2. Observation record organization: Combine multiple excrement images to assist in organizing visible features and observation records.
|
|
2363
|
+
* 3. Custom image-text analysis: Provide a user prompt to specify the image content to focus on and the analysis requirements.
|
|
2364
|
+
*
|
|
2365
|
+
* @remarks
|
|
2366
|
+
* 1. Multi-image input: image_url is a required array of HTTPS image URLs.
|
|
2367
|
+
* 2. Default excrement analysis: If user_prompt is not specified or is an empty string, the system determines whether excrement is present and outputs the location, urine clump/feces category, and subcategory for each region.
|
|
2368
|
+
* 3. Custom analysis: user_prompt and system_prompt are both optional. You can specify analysis requirements and system prompts separately. When submitting custom prompts, use values that comply with JSON string escaping rules.
|
|
2369
|
+
* 4. Text result response: result is always a String. Even if the model text itself is JSON, it is not automatically converted to an object.
|
|
2370
|
+
*
|
|
2371
|
+
* @param tmpReq - PetHealthAnalysisRequest
|
|
2372
|
+
* @param runtime - runtime options for this request RuntimeOptions
|
|
2373
|
+
* @returns PetHealthAnalysisResponse
|
|
2374
|
+
*/
|
|
2375
|
+
async petHealthAnalysisWithOptions(tmpReq, runtime) {
|
|
2376
|
+
tmpReq.validate();
|
|
2377
|
+
let request = new $_model.PetHealthAnalysisShrinkRequest({});
|
|
2378
|
+
openapi_core_2.OpenApiUtil.convert(tmpReq, request);
|
|
2379
|
+
if (!$dara.isNull(tmpReq.imageUrl)) {
|
|
2380
|
+
request.imageUrlShrink = openapi_core_2.OpenApiUtil.arrayToStringWithSpecifiedStyle(tmpReq.imageUrl, "ImageUrl", "json");
|
|
2381
|
+
}
|
|
2382
|
+
let query = {};
|
|
2383
|
+
if (!$dara.isNull(request.imageUrlShrink)) {
|
|
2384
|
+
query["ImageUrl"] = request.imageUrlShrink;
|
|
2385
|
+
}
|
|
2386
|
+
if (!$dara.isNull(request.systemPrompt)) {
|
|
2387
|
+
query["SystemPrompt"] = request.systemPrompt;
|
|
2388
|
+
}
|
|
2389
|
+
if (!$dara.isNull(request.userPrompt)) {
|
|
2390
|
+
query["UserPrompt"] = request.userPrompt;
|
|
2391
|
+
}
|
|
2392
|
+
let req = new openapi_core_2.$OpenApiUtil.OpenApiRequest({
|
|
2393
|
+
query: openapi_core_2.OpenApiUtil.query(query),
|
|
2394
|
+
});
|
|
2395
|
+
let params = new openapi_core_2.$OpenApiUtil.Params({
|
|
2396
|
+
action: "PetHealthAnalysis",
|
|
2397
|
+
version: "2026-04-28",
|
|
2398
|
+
protocol: "HTTPS",
|
|
2399
|
+
pathname: "/",
|
|
2400
|
+
method: "POST",
|
|
2401
|
+
authType: "AK",
|
|
2402
|
+
style: "RPC",
|
|
2403
|
+
reqBodyType: "formData",
|
|
2404
|
+
bodyType: "json",
|
|
2405
|
+
});
|
|
2406
|
+
return $dara.cast(await this.callApi(params, req, runtime), new $_model.PetHealthAnalysisResponse({}));
|
|
2407
|
+
}
|
|
2408
|
+
/**
|
|
2409
|
+
* Analyzes pet excrement images using a pet multimodal model building product. Supports uploading pet excrement images, determines by default whether excrement is present in the image, locates and performs categorization on each visible excrement area, and returns model-generated text containing the categorization results. The invoker can also optionally provide a user prompt and a system prompt to customize image analysis requirements and response format.
|
|
2410
|
+
* Applicable scenarios:
|
|
2411
|
+
* 1. Excrement image analysis: Upload pet excrement images for analysis using the default method.
|
|
2412
|
+
* 2. Observation record organization: Combine multiple excrement images to assist in organizing visible features and observation records.
|
|
2413
|
+
* 3. Custom image-text analysis: Provide a user prompt to specify the image content to focus on and the analysis requirements.
|
|
2414
|
+
*
|
|
2415
|
+
* @remarks
|
|
2416
|
+
* 1. Multi-image input: image_url is a required array of HTTPS image URLs.
|
|
2417
|
+
* 2. Default excrement analysis: If user_prompt is not specified or is an empty string, the system determines whether excrement is present and outputs the location, urine clump/feces category, and subcategory for each region.
|
|
2418
|
+
* 3. Custom analysis: user_prompt and system_prompt are both optional. You can specify analysis requirements and system prompts separately. When submitting custom prompts, use values that comply with JSON string escaping rules.
|
|
2419
|
+
* 4. Text result response: result is always a String. Even if the model text itself is JSON, it is not automatically converted to an object.
|
|
2420
|
+
*
|
|
2421
|
+
* @param request - PetHealthAnalysisRequest
|
|
2422
|
+
* @returns PetHealthAnalysisResponse
|
|
2423
|
+
*/
|
|
2424
|
+
async petHealthAnalysis(request) {
|
|
2425
|
+
let runtime = new $dara.RuntimeOptions({});
|
|
2426
|
+
return await this.petHealthAnalysisWithOptions(request, runtime);
|
|
2427
|
+
}
|
|
2101
2428
|
/**
|
|
2102
2429
|
* Maas-MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs, continuously optimized for e-commerce-specific terminology such as brands, models, materials, and attributes. It supports mutual translation among more than 400 language directions (including bridged translations), especially Chinese-English, Chinese-to-multilingual, and English-to-multilingual translations. It supports language detection for 24 language directions. If you do not specify a source language, automatic detection is performed.
|
|
2103
2430
|
*
|
|
@@ -2819,32 +3146,32 @@ class Client extends openapi_core_1.default {
|
|
|
2819
3146
|
return await this.textCorrectWithOptions(request, runtime);
|
|
2820
3147
|
}
|
|
2821
3148
|
/**
|
|
2822
|
-
* Maas-MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs,
|
|
3149
|
+
* Maas-MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs, with continuous optimization for e-commerce-specific terminology such as brands, models, materials, and attributes. It supports mutual translation among more than 400 language directions (including bridged directions), especially Chinese-English, Chinese-multilingual, and English-multilingual translation. It supports language detection for 31 languages in query phrases and 35 languages in general scenarios. If you do not specify a source language, automatic language detection is performed.
|
|
2823
3150
|
*
|
|
2824
3151
|
* @remarks
|
|
2825
3152
|
* ## Product Introduction
|
|
2826
|
-
* Marco MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs,
|
|
3153
|
+
* Marco MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs, with continuous optimization for e-commerce-specific terminology such as brands, models, materials, and attributes. It supports mutual translation among more than 100 language directions (including bridged directions), especially Chinese-English, Chinese-multilingual, and English-multilingual translation. It supports language detection for 24 languages. If you do not specify a source language, automatic language detection is performed.
|
|
2827
3154
|
* ## Common scenarios
|
|
2828
|
-
* -
|
|
2829
|
-
* -
|
|
2830
|
-
* -
|
|
2831
|
-
* -
|
|
2832
|
-
* Marco MT can also be used for general translation scenarios such as office work, meetings, and daily conversations.
|
|
3155
|
+
* - Product titles: Marco MT large model translation accurately translates phrases, keywords, and trending terms in product titles into the target language, improving product visibility on e-commerce platforms.
|
|
3156
|
+
* - Product descriptions: Product descriptions are typically long with diverse and complex formatting. Marco MT large model translation supports long text translation and HTML formatting, delivering excellent translation quality and format preservation.
|
|
3157
|
+
* - Product attributes: Product attribute terms are typically short and may contain e-commerce-specific terms such as model numbers and specifications.
|
|
3158
|
+
* - Customer service conversations: In customer service scenarios, Marco MT optimizes issues such as missing context and typos, and supports polite expressions for certain language directions.
|
|
3159
|
+
* Marco MT large model translation can also be used for general-purpose translation scenarios such as office work, meetings, and daily conversations.
|
|
2833
3160
|
* ## Functions and features
|
|
2834
3161
|
* - **Supported language directions**
|
|
2835
|
-
* Marco MT supports mutual translation among more than 100 languages and language detection for 24 languages. For supported language directions, see [Language direction mapping table](https://www.alibabacloud.com/help/en/document_detail/3041883.html).
|
|
3162
|
+
* Marco MT large model translation supports mutual translation among more than 100 languages and language detection for 24 languages. For supported language directions, see [Language direction mapping table](https://www.alibabacloud.com/help/en/document_detail/3041883.html).
|
|
2836
3163
|
* - **Ultra-low hallucination rate and high-speed inference**
|
|
2837
|
-
* Marco MT is a large language model with preference alignment for translation scenarios. Compared with general-purpose large language models, it
|
|
3164
|
+
* Marco MT large model translation is a large language model with preference alignment for translation scenarios. Compared with general-purpose large language models, it provides faster translation inference and an extremely low hallucination rate, making it especially suitable for large-scale invocation scenarios.
|
|
2838
3165
|
* - **Supported text format types**
|
|
2839
|
-
* Currently
|
|
2840
|
-
* - **Translation intervention**
|
|
2841
|
-
*
|
|
3166
|
+
* Currently, text and html text types are supported.
|
|
3167
|
+
* - **Translation intervention support**
|
|
3168
|
+
* Custom translation results are supported, including do-not-translate (ABC→ABC), specified translation (ABC→DEF), and skip translation (ABC→empty value). This is commonly used for scenarios such as brand name protection. Simply pass the corresponding intervention glossary ID when calling the API to meet your translation needs across different scenarios. You can upload up to 100,000 intervention terms. If you need more, contact the platform for assistance.
|
|
2842
3169
|
* ## Translation quality comparison
|
|
2843
3170
|
* | **Source text** | **Aidge translation** | **Other translation product 1** | **Other translation product 2** | **Aidge translation advantage** |
|
|
2844
3171
|
* |----------|---------------|-----------------------|-----------------------|---------------------|
|
|
2845
3172
|
* | 75mm 4Pin 0.38A GA81S2U 12V GTX 960 Cooling Fan for Zotac GTX960 Graphics Card Cooling Fan | 75mm 4Pin 0.38A GA81S2U 12V GTX 960, ventilador de enfriamiento para tarjeta gráfica Zotac GTX960. | Ventilador de refrigeración de 75mm 4Pin 0.38A GA81S2U 12V para la tarjeta gráfica Zotac GTX960 Ventilador de enfriamiento GTX 960. | Ventilador de refrigeración para tarjeta gráfica Zotac GTX960, 75mm, 4 pines, 0,38A, GA81S2U, 12V, GTX 960 | Clearer product subject |
|
|
2846
3173
|
* | New Arrival Classic Style Luxury Brand 6 Hands Swiss Automatic Mechanical Men\\"s Business Steel Watch | Nueva llegada, reloj mecánico automático de lujo con 6 manecillas, estilo clásico, de acero para hombres de negocios | Nueva Llegada Reloj De Acero de Negocios para Hombre con Mecanismo Automático Suizo y 6 Agujas de Estilo Clásico de Marca de Lujo. | Reloj de acero de negocios para hombre, mecánico automático suizo, marca de lujo, estilo clásico, 6 manos, nueva llegada | More complete sentence structure, clearer expression, and more reasonable segmentation |
|
|
2847
|
-
* | Men Key Bag Genuine Cow Leather Buckets Key Cases Pouch Zipper Keychain Auto Car Key Case Women Home Key Holder Wallet | Sac à clés pour hommes en cuir de vache véritable, étuis à clés, pochette à fermeture éclair, porte-clés, étui à clé de voiture, porte-clés pour femmes, organisateur de portefeuille | Porte-clés sac en cuir de vache véritable, étuis à clés en forme de seau, pochette à fermeture éclair, porte-clés auto pour voiture, étui à clés pour femmes, organisateur de clés de maison, portefeuille. | Hommes porte-clés en cuir de vache véritable sac/pochette fermature éclair porte-clés Auto voiture porte-clés femmes maison porte-clés organisateur portefeuille | Clearer and more accurate product subject expression
|
|
3174
|
+
* | Men Key Bag Genuine Cow Leather Buckets Key Cases Pouch Zipper Keychain Auto Car Key Case Women Home Key Holder Wallet | Sac à clés pour hommes en cuir de vache véritable, étuis à clés, pochette à fermeture éclair, porte-clés, étui à clé de voiture, porte-clés pour femmes, organisateur de portefeuille | Porte-clés sac en cuir de vache véritable, étuis à clés en forme de seau, pochette à fermeture éclair, porte-clés auto pour voiture, étui à clés pour femmes, organisateur de clés de maison, portefeuille. | Hommes porte-clés en cuir de vache véritable sac/pochette fermature éclair porte-clés Auto voiture porte-clés femmes maison porte-clés organisateur portefeuille | Clearer and more accurate product subject expression as "key bag" rather than "keychain", avoiding ambiguity about the product subject in other translations |
|
|
2848
3175
|
*
|
|
2849
3176
|
* @param tmpReq - TextTranslateRequest
|
|
2850
3177
|
* @param runtime - runtime options for this request RuntimeOptions
|
|
@@ -2896,32 +3223,32 @@ class Client extends openapi_core_1.default {
|
|
|
2896
3223
|
return $dara.cast(await this.callApi(params, req, runtime), new $_model.TextTranslateResponse({}));
|
|
2897
3224
|
}
|
|
2898
3225
|
/**
|
|
2899
|
-
* Maas-MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs,
|
|
3226
|
+
* Maas-MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs, with continuous optimization for e-commerce-specific terminology such as brands, models, materials, and attributes. It supports mutual translation among more than 400 language directions (including bridged directions), especially Chinese-English, Chinese-multilingual, and English-multilingual translation. It supports language detection for 31 languages in query phrases and 35 languages in general scenarios. If you do not specify a source language, automatic language detection is performed.
|
|
2900
3227
|
*
|
|
2901
3228
|
* @remarks
|
|
2902
3229
|
* ## Product Introduction
|
|
2903
|
-
* Marco MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs,
|
|
3230
|
+
* Marco MT is an e-commerce translation model trained on over 1 billion high-quality bilingual sentence pairs, with continuous optimization for e-commerce-specific terminology such as brands, models, materials, and attributes. It supports mutual translation among more than 100 language directions (including bridged directions), especially Chinese-English, Chinese-multilingual, and English-multilingual translation. It supports language detection for 24 languages. If you do not specify a source language, automatic language detection is performed.
|
|
2904
3231
|
* ## Common scenarios
|
|
2905
|
-
* -
|
|
2906
|
-
* -
|
|
2907
|
-
* -
|
|
2908
|
-
* -
|
|
2909
|
-
* Marco MT can also be used for general translation scenarios such as office work, meetings, and daily conversations.
|
|
3232
|
+
* - Product titles: Marco MT large model translation accurately translates phrases, keywords, and trending terms in product titles into the target language, improving product visibility on e-commerce platforms.
|
|
3233
|
+
* - Product descriptions: Product descriptions are typically long with diverse and complex formatting. Marco MT large model translation supports long text translation and HTML formatting, delivering excellent translation quality and format preservation.
|
|
3234
|
+
* - Product attributes: Product attribute terms are typically short and may contain e-commerce-specific terms such as model numbers and specifications.
|
|
3235
|
+
* - Customer service conversations: In customer service scenarios, Marco MT optimizes issues such as missing context and typos, and supports polite expressions for certain language directions.
|
|
3236
|
+
* Marco MT large model translation can also be used for general-purpose translation scenarios such as office work, meetings, and daily conversations.
|
|
2910
3237
|
* ## Functions and features
|
|
2911
3238
|
* - **Supported language directions**
|
|
2912
|
-
* Marco MT supports mutual translation among more than 100 languages and language detection for 24 languages. For supported language directions, see [Language direction mapping table](https://www.alibabacloud.com/help/en/document_detail/3041883.html).
|
|
3239
|
+
* Marco MT large model translation supports mutual translation among more than 100 languages and language detection for 24 languages. For supported language directions, see [Language direction mapping table](https://www.alibabacloud.com/help/en/document_detail/3041883.html).
|
|
2913
3240
|
* - **Ultra-low hallucination rate and high-speed inference**
|
|
2914
|
-
* Marco MT is a large language model with preference alignment for translation scenarios. Compared with general-purpose large language models, it
|
|
3241
|
+
* Marco MT large model translation is a large language model with preference alignment for translation scenarios. Compared with general-purpose large language models, it provides faster translation inference and an extremely low hallucination rate, making it especially suitable for large-scale invocation scenarios.
|
|
2915
3242
|
* - **Supported text format types**
|
|
2916
|
-
* Currently
|
|
2917
|
-
* - **Translation intervention**
|
|
2918
|
-
*
|
|
3243
|
+
* Currently, text and html text types are supported.
|
|
3244
|
+
* - **Translation intervention support**
|
|
3245
|
+
* Custom translation results are supported, including do-not-translate (ABC→ABC), specified translation (ABC→DEF), and skip translation (ABC→empty value). This is commonly used for scenarios such as brand name protection. Simply pass the corresponding intervention glossary ID when calling the API to meet your translation needs across different scenarios. You can upload up to 100,000 intervention terms. If you need more, contact the platform for assistance.
|
|
2919
3246
|
* ## Translation quality comparison
|
|
2920
3247
|
* | **Source text** | **Aidge translation** | **Other translation product 1** | **Other translation product 2** | **Aidge translation advantage** |
|
|
2921
3248
|
* |----------|---------------|-----------------------|-----------------------|---------------------|
|
|
2922
3249
|
* | 75mm 4Pin 0.38A GA81S2U 12V GTX 960 Cooling Fan for Zotac GTX960 Graphics Card Cooling Fan | 75mm 4Pin 0.38A GA81S2U 12V GTX 960, ventilador de enfriamiento para tarjeta gráfica Zotac GTX960. | Ventilador de refrigeración de 75mm 4Pin 0.38A GA81S2U 12V para la tarjeta gráfica Zotac GTX960 Ventilador de enfriamiento GTX 960. | Ventilador de refrigeración para tarjeta gráfica Zotac GTX960, 75mm, 4 pines, 0,38A, GA81S2U, 12V, GTX 960 | Clearer product subject |
|
|
2923
3250
|
* | New Arrival Classic Style Luxury Brand 6 Hands Swiss Automatic Mechanical Men\\"s Business Steel Watch | Nueva llegada, reloj mecánico automático de lujo con 6 manecillas, estilo clásico, de acero para hombres de negocios | Nueva Llegada Reloj De Acero de Negocios para Hombre con Mecanismo Automático Suizo y 6 Agujas de Estilo Clásico de Marca de Lujo. | Reloj de acero de negocios para hombre, mecánico automático suizo, marca de lujo, estilo clásico, 6 manos, nueva llegada | More complete sentence structure, clearer expression, and more reasonable segmentation |
|
|
2924
|
-
* | Men Key Bag Genuine Cow Leather Buckets Key Cases Pouch Zipper Keychain Auto Car Key Case Women Home Key Holder Wallet | Sac à clés pour hommes en cuir de vache véritable, étuis à clés, pochette à fermeture éclair, porte-clés, étui à clé de voiture, porte-clés pour femmes, organisateur de portefeuille | Porte-clés sac en cuir de vache véritable, étuis à clés en forme de seau, pochette à fermeture éclair, porte-clés auto pour voiture, étui à clés pour femmes, organisateur de clés de maison, portefeuille. | Hommes porte-clés en cuir de vache véritable sac/pochette fermature éclair porte-clés Auto voiture porte-clés femmes maison porte-clés organisateur portefeuille | Clearer and more accurate product subject expression
|
|
3251
|
+
* | Men Key Bag Genuine Cow Leather Buckets Key Cases Pouch Zipper Keychain Auto Car Key Case Women Home Key Holder Wallet | Sac à clés pour hommes en cuir de vache véritable, étuis à clés, pochette à fermeture éclair, porte-clés, étui à clé de voiture, porte-clés pour femmes, organisateur de portefeuille | Porte-clés sac en cuir de vache véritable, étuis à clés en forme de seau, pochette à fermeture éclair, porte-clés auto pour voiture, étui à clés pour femmes, organisateur de clés de maison, portefeuille. | Hommes porte-clés en cuir de vache véritable sac/pochette fermature éclair porte-clés Auto voiture porte-clés femmes maison porte-clés organisateur portefeuille | Clearer and more accurate product subject expression as "key bag" rather than "keychain", avoiding ambiguity about the product subject in other translations |
|
|
2925
3252
|
*
|
|
2926
3253
|
* @param request - TextTranslateRequest
|
|
2927
3254
|
* @returns TextTranslateResponse
|