@alicloud/aidge20260428 5.12.0 → 5.12.2

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 CHANGED
@@ -1961,23 +1961,21 @@ export default class Client extends OpenApi {
1961
1961
  */
1962
1962
  imageTranslationStandardAsync(request: $_model.ImageTranslationStandardAsyncRequest): Promise<$_model.ImageTranslationStandardAsyncResponse>;
1963
1963
  /**
1964
- * Synchronously upscales e-commerce images to intelligently enhance resolution and details by improving textures and optimizing image quality. Supports high-definition outputs such as 2K and 4K to meet the requirements for main product images, detail images, and marketing materials, enabling efficient processing of large image batches to improve overall product image quality.
1964
+ * Upscales e-commerce images by intelligently enhancing resolution and details, improving textures and image quality during the upscaling process. Supports high-definition outputs such as 2K and 4K to meet the requirements for product main images, detail images, and marketing materials, and enables efficient processing of large batches of images to improve overall image quality. (Synchronous)
1965
1965
  *
1966
1966
  * @remarks
1967
1967
  * # Product introduction
1968
- * The high-definition upscaling product is tailored for e-commerce images. It intelligently enhances image resolution and details, improves textures and image quality during upscaling, and supports high-definition outputs such as 2K and 4K. It meets the high-definition requirements for product main images, detail images, and marketing materials, enabling efficient and clear processing of large batches of images to improve product image quality. (Synchronous)
1968
+ * The high-definition upscaling product is customized for e-commerce images. It intelligently enhances image resolution and details, improves textures and image quality during the upscaling process, and supports high-definition outputs such as 2K and 4K. It meets the high-definition requirements for product main images, detail images, and marketing materials, enabling efficient and clear processing of large batches of images to improve product image quality. (Synchronous)
1969
1969
  * # Scenarios
1970
1970
  * * High-definition e-commerce product images: Enhances the resolution and details of product main images, detail images, and marketing materials to make textures and materials clearer and improve product display quality.
1971
- *
1972
- * * Multi-platform product material adaptation: Upscales existing product images in high definition to meet the image display requirements of different e-commerce platforms and adapt to multi-platform and multi-scenario material usage needs.
1973
- *
1974
- * * Batch optimization of existing images: Enhances existing low-resolution product images in high definition to reduce reshooting and production costs and improve the reuse efficiency of image materials.
1971
+ * * Multi-platform product material adaptation: Upscales existing product images to high definition to meet the image display requirements of different e-commerce platforms and adapt to multi-platform and multi-scenario material usage needs.
1972
+ * * Batch optimization of existing images: Enhances existing low-resolution product images to high definition, reducing the costs of reshooting and remaking, and improving the reuse efficiency of image materials.
1975
1973
  * # Features
1976
- * 1. Intelligent high-definition upscaling: Uses AI-based image enhancement technology to improve the resolution and optimize the details of input images, enhancing image clarity and texture representation.
1974
+ * 1. Intelligent high-definition upscaling: Based on AI image enhancement technology, it improves the resolution and optimizes the details of input images to enhance image clarity and texture representation.
1977
1975
  *
1978
- * 2. Multiple upscaling factors: Supports 2x and 4x upscaling factors. You can flexibly select the factor based on image quality and business requirements.
1976
+ * 2. Multiple upscaling factors: Supports 2x and 4x upscaling factors, allowing flexible selection based on image quality and business requirements.
1979
1977
  *
1980
- * 3. Synchronous API calls: Supports synchronous submission of images and retrieval of processing results through API calls, facilitating integration into e-commerce image processing workflows.
1978
+ * 3. Synchronous API calls: Supports synchronous submission of images and retrieval of processing results through APIs, facilitating integration into e-commerce image processing workflows.
1981
1979
  * # Effect comparison
1982
1980
  * | **Original image** | **Upscaled image** |
1983
1981
  * | --- | --- |
@@ -1989,23 +1987,21 @@ export default class Client extends OpenApi {
1989
1987
  */
1990
1988
  imageUpscaleWithOptions(request: $_model.ImageUpscaleRequest, runtime: $dara.RuntimeOptions): Promise<$_model.ImageUpscaleResponse>;
1991
1989
  /**
1992
- * Synchronously upscales e-commerce images to intelligently enhance resolution and details by improving textures and optimizing image quality. Supports high-definition outputs such as 2K and 4K to meet the requirements for main product images, detail images, and marketing materials, enabling efficient processing of large image batches to improve overall product image quality.
1990
+ * Upscales e-commerce images by intelligently enhancing resolution and details, improving textures and image quality during the upscaling process. Supports high-definition outputs such as 2K and 4K to meet the requirements for product main images, detail images, and marketing materials, and enables efficient processing of large batches of images to improve overall image quality. (Synchronous)
1993
1991
  *
1994
1992
  * @remarks
1995
1993
  * # Product introduction
1996
- * The high-definition upscaling product is tailored for e-commerce images. It intelligently enhances image resolution and details, improves textures and image quality during upscaling, and supports high-definition outputs such as 2K and 4K. It meets the high-definition requirements for product main images, detail images, and marketing materials, enabling efficient and clear processing of large batches of images to improve product image quality. (Synchronous)
1994
+ * The high-definition upscaling product is customized for e-commerce images. It intelligently enhances image resolution and details, improves textures and image quality during the upscaling process, and supports high-definition outputs such as 2K and 4K. It meets the high-definition requirements for product main images, detail images, and marketing materials, enabling efficient and clear processing of large batches of images to improve product image quality. (Synchronous)
1997
1995
  * # Scenarios
1998
1996
  * * High-definition e-commerce product images: Enhances the resolution and details of product main images, detail images, and marketing materials to make textures and materials clearer and improve product display quality.
1999
- *
2000
- * * Multi-platform product material adaptation: Upscales existing product images in high definition to meet the image display requirements of different e-commerce platforms and adapt to multi-platform and multi-scenario material usage needs.
2001
- *
2002
- * * Batch optimization of existing images: Enhances existing low-resolution product images in high definition to reduce reshooting and production costs and improve the reuse efficiency of image materials.
1997
+ * * Multi-platform product material adaptation: Upscales existing product images to high definition to meet the image display requirements of different e-commerce platforms and adapt to multi-platform and multi-scenario material usage needs.
1998
+ * * Batch optimization of existing images: Enhances existing low-resolution product images to high definition, reducing the costs of reshooting and remaking, and improving the reuse efficiency of image materials.
2003
1999
  * # Features
2004
- * 1. Intelligent high-definition upscaling: Uses AI-based image enhancement technology to improve the resolution and optimize the details of input images, enhancing image clarity and texture representation.
2000
+ * 1. Intelligent high-definition upscaling: Based on AI image enhancement technology, it improves the resolution and optimizes the details of input images to enhance image clarity and texture representation.
2005
2001
  *
2006
- * 2. Multiple upscaling factors: Supports 2x and 4x upscaling factors. You can flexibly select the factor based on image quality and business requirements.
2002
+ * 2. Multiple upscaling factors: Supports 2x and 4x upscaling factors, allowing flexible selection based on image quality and business requirements.
2007
2003
  *
2008
- * 3. Synchronous API calls: Supports synchronous submission of images and retrieval of processing results through API calls, facilitating integration into e-commerce image processing workflows.
2004
+ * 3. Synchronous API calls: Supports synchronous submission of images and retrieval of processing results through APIs, facilitating integration into e-commerce image processing workflows.
2009
2005
  * # Effect comparison
2010
2006
  * | **Original image** | **Upscaled image** |
2011
2007
  * | --- | --- |
@@ -3557,61 +3553,61 @@ export default class Client extends OpenApi {
3557
3553
  */
3558
3554
  virtualTryOn(request: $_model.VirtualTryOnRequest): Promise<$_model.VirtualTryOnResponse>;
3559
3555
  /**
3560
- * Provides a one-stop AI-powered image processing service for e-commerce sellers. Orchestrates seven atomic capabilities — element detection, intelligent matting, intelligent removal, Image Translation Pro, image expansion, intelligent cropping, and HD upscaling — into an image processing workflow. Upload an image once, select the desired capabilities, and complete multiple image optimizations sequentially in a single call to produce product images that meet listing platform requirements. (Asynchronous)
3556
+ * Provides a one-stop AI image processing service for e-commerce sellers. Orchestrates six atomic capabilities—element recognition, intelligent cutout, intelligent erasure, image translation Pro, intelligent cropping, and high-definition upscaling—into an image processing workflow. Users can upload an image once, select the required capabilities, and complete multiple image optimizations in sequence to output product images that meet the requirements of e-commerce platforms. (Asynchronous)
3561
3557
  *
3562
3558
  * @remarks
3563
- * # 1. Product Introduction
3564
- * A one-stop AI-powered image processing service for e-commerce sellers. Orchestrates seven atomic capabilities — element detection, intelligent matting, intelligent removal, Image Translation Pro, image expansion, intelligent cropping, and HD upscaling — into an image processing workflow. Upload an image once, select the desired capabilities, and complete multiple image optimizations sequentially in a single call to produce product images that meet listing platform requirements. (Asynchronous)
3565
- * # 2. Common scenarios
3566
- * Applicable to batch optimization of product images in cross-platform listing scenarios. Adapts source product images into main images, SKU images, and detail images required by listing platforms. Supported capabilities include:
3567
- * * Detecting low-quality images that contain text, watermarks, logos, or overlay patches, allowing users to filter images that need processing
3559
+ * # 1. Product introduction
3560
+ * Provides a one-stop AI image processing service for e-commerce sellers. Orchestrates six atomic capabilities—element recognition, intelligent cutout, intelligent erasure, image translation Pro, intelligent cropping, and high-definition upscaling—into an image processing workflow. Users can upload an image once, select the required capabilities, and complete multiple image optimizations in sequence to output product images that meet the requirements of e-commerce platforms. (Asynchronous)
3561
+ * # 2. Scenarios
3562
+ * Applicable to batch optimization of product images in cross-platform listing scenarios. Adapts source product images into main images, SKU images, and detail images required by listing platforms. Supports the following operations:
3563
+ * * Identifies low-quality images with text, watermarks, logos, or cluttered text, and allows users to filter the images to be processed.
3568
3564
  *
3569
- * * Removing non-compliant elements such as watermarks, logos, overlay patches, and Chinese text from original images
3565
+ * * Removes non-compliant elements such as watermarks, logos, cluttered text, and Chinese text from the original image.
3570
3566
  *
3571
- * * Removing the background around the product subject edges
3567
+ * * Removes the background around the main product subject.
3572
3568
  *
3573
- * * Translating text within images into other languages
3569
+ * * Translates text in the image into other languages.
3574
3570
  *
3575
- * * Expanding images to other dimensions (by ratio, size, etc.) with automatic background fill
3571
+ * * Expands the image to other sizes (by ratio or dimensions) and automatically fills the background.
3576
3572
  *
3577
- * * Cropping images to specified dimensions
3573
+ * * Crops the image to a specified size.
3578
3574
  *
3579
- * * Upscaling image resolution by a specified factor
3575
+ * * Upscales the image pixels by a specified factor.
3580
3576
  * # 3. Features
3581
- * Provide a product image URL and use the Ability parameter to select the desired combination of AI capabilities (1 = Intelligent Element Detection, 2 = Intelligent Matting, 3 = Intelligent Removal, 4 = Image Translation Pro, 5 = Image Expansion, 6 = Intelligent Cropping, 7 = HD Upscaling). The system executes the selected capabilities sequentially in a preset order. Because Intelligent Element Detection helps determine whether an image contains certain specified elements, and the presence of those elements affects the selection of subsequent AI capabilities, an additional input is provided that allows users to choose whether to filter images out of or retain images in the subsequent AI processing pipeline.
3577
+ * Users input the URL of a product image and select the required AI capabilities through the Ability parameter (1=Intelligent element recognition, 2=Intelligent cutout, 3=Intelligent erasure, 4=Image translation Pro, 5=Image expansion, 6=Intelligent cropping, 7=High-definition upscaling). The system executes the selected capabilities in a preset sequence. Because intelligent element recognition helps users determine whether an image contains specific elements, and the presence of these elements affects the selection of subsequent AI capabilities, an additional input option is provided. This option allows users to choose whether to filter out images from subsequent AI processing or retain them without further processing.
3582
3578
  * ![ai-pipeline-flowchart.svg](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/bddb522a-0795-4b5c-b25d-10e933426367.svg)
3583
- * Each AI capability is described as follows:
3584
- * 1. **Intelligent Element Detection:** An element detection capability designed specifically for e-commerce images. Detects elements such as text, logos, watermarks, and text-bearing color blocks in both the image subject and background, and returns Boolean detection results. Supports filtering images based on detection results (for example, processing only images that contain watermarks). Works in conjunction with subsequent capabilities — first identifies the types of elements that need optimization, then precisely executes removal and other operations.
3579
+ * The following section describes each AI capability:
3580
+ * **1. Intelligent element recognition**: An element detection capability designed specifically for e-commerce images. It identifies elements such as text, logos, watermarks, and text blocks in the subject and background of an image, and returns Boolean detection results. Users can filter images based on the recognition results (for example, processing only images with watermarks) and link this capability with subsequent ones—first identifying the types of elements to be optimized, and then precisely executing operations such as erasure.
3585
3581
  *
3586
- * 2. **Intelligent Matting:** Automatically identifies the product subject in an image and precisely separates the subject from the background. Based on deep learning models, supports accurate segmentation of complex edges such as hair strands and transparent objects. Outputs images with a transparent or white background, supports custom background colors and target size cropping, and directly generates product display images that comply with e-commerce platform specifications.
3582
+ * **2. Intelligent cutout**: Automatically identifies the main product subject in an image and precisely separates it from the background. Based on deep learning models, it supports precise segmentation of complex edges such as hair and transparent objects. It can output images with transparent or white backgrounds, supports custom background colors and target size cropping, and directly generates product display images that comply with e-commerce platform specifications.
3587
3583
  *
3588
- * 3. **Intelligent Removal:** Automatically detects and removes non-compliant elements from e-commerce images, including text, specific names (brand/store names), transparent text blocks, and overlay patches. Supports specifying removal of elements in the subject area or non-subject area separately. Also supports specifying exact removal regions through the Mask parameter for efficient compliance cleanup of large batches of images.
3584
+ * **3. Intelligent erasure**: Automatically identifies and removes non-compliant elements in e-commerce images, such as text, specific names (brand or store names), transparent text blocks, and cluttered text. It supports specifying the removal of elements in the subject area or non-subject area separately. It also supports precisely specifying the removal area through the Mask parameter to achieve efficient and compliant cleanup of large batches of images.
3589
3585
  *
3590
- * 4. **Image Translation Pro:** Leverages multimodal foundation model technology to translate text in images into a target language and render it directly at the corresponding position in the original image. Supports more than 100 language pairs. Provides product subject text protection, brand name protection, and translation intervention glossaries to prevent mistranslation of embedded product information. After translation, returns layout information such as text position, font, and color to support secondary editing.
3586
+ * **4. Image translation Pro**: Integrates multimodal large model technology to translate text in an image into a target language and directly renders it in the corresponding position of the original image. It supports over 100 language pairs. It provides features such as product subject text protection, brand name protection, and translation intervention glossaries to avoid mistranslating embedded product information. After translation, it can return layout information such as text position, font, and color to support secondary editing.
3591
3587
  *
3592
- * 5. **Intelligent Image Expansion:** Expands the image canvas using AI generation technology to broaden the image view without cropping the subject. Supports multiple expansion methods: by aspect ratio (1:1, 3:4, 4:3, 9:16, 16:9), by horizontal/vertical scaling ratio (1.0× to 3.0×), or by pixel offset in the up, down, left, and right directions. AI automatically fills in naturally transitioning background content.
3588
+ * **5. Intelligent image expansion**: Expands the image canvas based on AI generation technology to broaden the image field of view without cropping the subject. It supports multiple expansion methods: expanding by aspect ratio (1:1, 3:4, 4:3, 9:16, 16:9), scaling by horizontal or vertical proportion (1.0 to 3.0 times), and shifting pixels in the up, down, left, or right directions. The AI automatically fills and generates naturally transitioning background content.
3593
3589
  *
3594
- * 6. **Intelligent Cropping:** Intelligently crops images to specified dimensions, automatically identifies the subject area, and precisely preserves core content. Supports custom output width and height (100–5000 pixels) to meet the image dimension requirements of different e-commerce platforms and ad placements, ensuring the subject remains prominent and the composition stays balanced after cropping.
3590
+ * **6. Intelligent cropping**: Intelligently crops images to specified sizes, automatically identifies the main subject area of the image, and precisely retains the core content. It supports custom output width and height (100 to 5000 pixels) and adapts to the image size requirements of different e-commerce platforms and ad slots, ensuring that the subject stands out and the picture is balanced after cropping.
3595
3591
  *
3596
- * 7. **HD Upscaling:** Enhances details and reduces noise while upscaling images. Supports 2× to 4× upscaling. Suitable for improving the clarity of low-resolution product images and optimizing visual presentation. Supports output in PNG, JPG, BMP, and other formats.
3597
- * # 4. Before and after comparison
3592
+ * **7. High-definition upscaling**: Enhances details and reduces noise while upscaling the image. It supports 2x to 4x upscaling. It is applicable to improving the clarity of low-resolution product images and optimizing visual presentation. It supports output in formats such as PNG, JPG, and BMP.
3593
+ * # 4. Effect comparison
3598
3594
  * | **Original image** | **Capabilities used** | **Output image** |
3599
3595
  * | --- | --- | --- |
3600
- * | ![image.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/3c1af966-8a5a-491d-9bfa-aefc321c9bed.png) | * Intelligent detection
3596
+ * | ![image.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/3c1af966-8a5a-491d-9bfa-aefc321c9bed.png) | * Intelligent recognition
3601
3597
  *
3602
- * * Image Translation Pro
3598
+ * * Image translation Pro
3603
3599
  *
3604
3600
  * * Intelligent cropping | ![R1.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/c2579e3e-389e-4717-84e3-834e60fea6b0.png) |
3605
- * | ![image.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/a92d14a9-ecf4-4e77-ad96-a2fcdd95dd1d.png) | * Intelligent matting
3601
+ * | ![image.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/a92d14a9-ecf4-4e77-ad96-a2fcdd95dd1d.png) | * Intelligent cutout
3606
3602
  *
3607
- * * Intelligent removal
3603
+ * * Intelligent erasure
3608
3604
  *
3609
3605
  * * Intelligent cropping | ![R1.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/f631f578-3496-4aa1-b5ef-cd1ea603e049.png) |
3610
- * | ![C2.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/12965cf8-cfdf-4a85-95fe-fdfcefcaa4ae.png) | * Intelligent detection
3606
+ * | ![C2.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/12965cf8-cfdf-4a85-95fe-fdfcefcaa4ae.png) | * Intelligent recognition
3611
3607
  *
3612
- * * Intelligent removal
3608
+ * * Intelligent erasure
3613
3609
  *
3614
- * * HD upscaling | ![R2.jpeg](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/95d29e79-8622-45fb-b284-cd580fe40ea0.jpeg) |
3610
+ * * High-definition upscaling | ![R2.jpeg](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/95d29e79-8622-45fb-b284-cd580fe40ea0.jpeg) |
3615
3611
  *
3616
3612
  * @param tmpReq - VisionFlowRequest
3617
3613
  * @param runtime - runtime options for this request RuntimeOptions
@@ -3619,61 +3615,61 @@ export default class Client extends OpenApi {
3619
3615
  */
3620
3616
  visionFlowWithOptions(tmpReq: $_model.VisionFlowRequest, runtime: $dara.RuntimeOptions): Promise<$_model.VisionFlowResponse>;
3621
3617
  /**
3622
- * Provides a one-stop AI-powered image processing service for e-commerce sellers. Orchestrates seven atomic capabilities — element detection, intelligent matting, intelligent removal, Image Translation Pro, image expansion, intelligent cropping, and HD upscaling — into an image processing workflow. Upload an image once, select the desired capabilities, and complete multiple image optimizations sequentially in a single call to produce product images that meet listing platform requirements. (Asynchronous)
3618
+ * Provides a one-stop AI image processing service for e-commerce sellers. Orchestrates six atomic capabilities—element recognition, intelligent cutout, intelligent erasure, image translation Pro, intelligent cropping, and high-definition upscaling—into an image processing workflow. Users can upload an image once, select the required capabilities, and complete multiple image optimizations in sequence to output product images that meet the requirements of e-commerce platforms. (Asynchronous)
3623
3619
  *
3624
3620
  * @remarks
3625
- * # 1. Product Introduction
3626
- * A one-stop AI-powered image processing service for e-commerce sellers. Orchestrates seven atomic capabilities — element detection, intelligent matting, intelligent removal, Image Translation Pro, image expansion, intelligent cropping, and HD upscaling — into an image processing workflow. Upload an image once, select the desired capabilities, and complete multiple image optimizations sequentially in a single call to produce product images that meet listing platform requirements. (Asynchronous)
3627
- * # 2. Common scenarios
3628
- * Applicable to batch optimization of product images in cross-platform listing scenarios. Adapts source product images into main images, SKU images, and detail images required by listing platforms. Supported capabilities include:
3629
- * * Detecting low-quality images that contain text, watermarks, logos, or overlay patches, allowing users to filter images that need processing
3621
+ * # 1. Product introduction
3622
+ * Provides a one-stop AI image processing service for e-commerce sellers. Orchestrates six atomic capabilities—element recognition, intelligent cutout, intelligent erasure, image translation Pro, intelligent cropping, and high-definition upscaling—into an image processing workflow. Users can upload an image once, select the required capabilities, and complete multiple image optimizations in sequence to output product images that meet the requirements of e-commerce platforms. (Asynchronous)
3623
+ * # 2. Scenarios
3624
+ * Applicable to batch optimization of product images in cross-platform listing scenarios. Adapts source product images into main images, SKU images, and detail images required by listing platforms. Supports the following operations:
3625
+ * * Identifies low-quality images with text, watermarks, logos, or cluttered text, and allows users to filter the images to be processed.
3630
3626
  *
3631
- * * Removing non-compliant elements such as watermarks, logos, overlay patches, and Chinese text from original images
3627
+ * * Removes non-compliant elements such as watermarks, logos, cluttered text, and Chinese text from the original image.
3632
3628
  *
3633
- * * Removing the background around the product subject edges
3629
+ * * Removes the background around the main product subject.
3634
3630
  *
3635
- * * Translating text within images into other languages
3631
+ * * Translates text in the image into other languages.
3636
3632
  *
3637
- * * Expanding images to other dimensions (by ratio, size, etc.) with automatic background fill
3633
+ * * Expands the image to other sizes (by ratio or dimensions) and automatically fills the background.
3638
3634
  *
3639
- * * Cropping images to specified dimensions
3635
+ * * Crops the image to a specified size.
3640
3636
  *
3641
- * * Upscaling image resolution by a specified factor
3637
+ * * Upscales the image pixels by a specified factor.
3642
3638
  * # 3. Features
3643
- * Provide a product image URL and use the Ability parameter to select the desired combination of AI capabilities (1 = Intelligent Element Detection, 2 = Intelligent Matting, 3 = Intelligent Removal, 4 = Image Translation Pro, 5 = Image Expansion, 6 = Intelligent Cropping, 7 = HD Upscaling). The system executes the selected capabilities sequentially in a preset order. Because Intelligent Element Detection helps determine whether an image contains certain specified elements, and the presence of those elements affects the selection of subsequent AI capabilities, an additional input is provided that allows users to choose whether to filter images out of or retain images in the subsequent AI processing pipeline.
3639
+ * Users input the URL of a product image and select the required AI capabilities through the Ability parameter (1=Intelligent element recognition, 2=Intelligent cutout, 3=Intelligent erasure, 4=Image translation Pro, 5=Image expansion, 6=Intelligent cropping, 7=High-definition upscaling). The system executes the selected capabilities in a preset sequence. Because intelligent element recognition helps users determine whether an image contains specific elements, and the presence of these elements affects the selection of subsequent AI capabilities, an additional input option is provided. This option allows users to choose whether to filter out images from subsequent AI processing or retain them without further processing.
3644
3640
  * ![ai-pipeline-flowchart.svg](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/bddb522a-0795-4b5c-b25d-10e933426367.svg)
3645
- * Each AI capability is described as follows:
3646
- * 1. **Intelligent Element Detection:** An element detection capability designed specifically for e-commerce images. Detects elements such as text, logos, watermarks, and text-bearing color blocks in both the image subject and background, and returns Boolean detection results. Supports filtering images based on detection results (for example, processing only images that contain watermarks). Works in conjunction with subsequent capabilities — first identifies the types of elements that need optimization, then precisely executes removal and other operations.
3641
+ * The following section describes each AI capability:
3642
+ * **1. Intelligent element recognition**: An element detection capability designed specifically for e-commerce images. It identifies elements such as text, logos, watermarks, and text blocks in the subject and background of an image, and returns Boolean detection results. Users can filter images based on the recognition results (for example, processing only images with watermarks) and link this capability with subsequent ones—first identifying the types of elements to be optimized, and then precisely executing operations such as erasure.
3647
3643
  *
3648
- * 2. **Intelligent Matting:** Automatically identifies the product subject in an image and precisely separates the subject from the background. Based on deep learning models, supports accurate segmentation of complex edges such as hair strands and transparent objects. Outputs images with a transparent or white background, supports custom background colors and target size cropping, and directly generates product display images that comply with e-commerce platform specifications.
3644
+ * **2. Intelligent cutout**: Automatically identifies the main product subject in an image and precisely separates it from the background. Based on deep learning models, it supports precise segmentation of complex edges such as hair and transparent objects. It can output images with transparent or white backgrounds, supports custom background colors and target size cropping, and directly generates product display images that comply with e-commerce platform specifications.
3649
3645
  *
3650
- * 3. **Intelligent Removal:** Automatically detects and removes non-compliant elements from e-commerce images, including text, specific names (brand/store names), transparent text blocks, and overlay patches. Supports specifying removal of elements in the subject area or non-subject area separately. Also supports specifying exact removal regions through the Mask parameter for efficient compliance cleanup of large batches of images.
3646
+ * **3. Intelligent erasure**: Automatically identifies and removes non-compliant elements in e-commerce images, such as text, specific names (brand or store names), transparent text blocks, and cluttered text. It supports specifying the removal of elements in the subject area or non-subject area separately. It also supports precisely specifying the removal area through the Mask parameter to achieve efficient and compliant cleanup of large batches of images.
3651
3647
  *
3652
- * 4. **Image Translation Pro:** Leverages multimodal foundation model technology to translate text in images into a target language and render it directly at the corresponding position in the original image. Supports more than 100 language pairs. Provides product subject text protection, brand name protection, and translation intervention glossaries to prevent mistranslation of embedded product information. After translation, returns layout information such as text position, font, and color to support secondary editing.
3648
+ * **4. Image translation Pro**: Integrates multimodal large model technology to translate text in an image into a target language and directly renders it in the corresponding position of the original image. It supports over 100 language pairs. It provides features such as product subject text protection, brand name protection, and translation intervention glossaries to avoid mistranslating embedded product information. After translation, it can return layout information such as text position, font, and color to support secondary editing.
3653
3649
  *
3654
- * 5. **Intelligent Image Expansion:** Expands the image canvas using AI generation technology to broaden the image view without cropping the subject. Supports multiple expansion methods: by aspect ratio (1:1, 3:4, 4:3, 9:16, 16:9), by horizontal/vertical scaling ratio (1.0× to 3.0×), or by pixel offset in the up, down, left, and right directions. AI automatically fills in naturally transitioning background content.
3650
+ * **5. Intelligent image expansion**: Expands the image canvas based on AI generation technology to broaden the image field of view without cropping the subject. It supports multiple expansion methods: expanding by aspect ratio (1:1, 3:4, 4:3, 9:16, 16:9), scaling by horizontal or vertical proportion (1.0 to 3.0 times), and shifting pixels in the up, down, left, or right directions. The AI automatically fills and generates naturally transitioning background content.
3655
3651
  *
3656
- * 6. **Intelligent Cropping:** Intelligently crops images to specified dimensions, automatically identifies the subject area, and precisely preserves core content. Supports custom output width and height (100–5000 pixels) to meet the image dimension requirements of different e-commerce platforms and ad placements, ensuring the subject remains prominent and the composition stays balanced after cropping.
3652
+ * **6. Intelligent cropping**: Intelligently crops images to specified sizes, automatically identifies the main subject area of the image, and precisely retains the core content. It supports custom output width and height (100 to 5000 pixels) and adapts to the image size requirements of different e-commerce platforms and ad slots, ensuring that the subject stands out and the picture is balanced after cropping.
3657
3653
  *
3658
- * 7. **HD Upscaling:** Enhances details and reduces noise while upscaling images. Supports 2× to 4× upscaling. Suitable for improving the clarity of low-resolution product images and optimizing visual presentation. Supports output in PNG, JPG, BMP, and other formats.
3659
- * # 4. Before and after comparison
3654
+ * **7. High-definition upscaling**: Enhances details and reduces noise while upscaling the image. It supports 2x to 4x upscaling. It is applicable to improving the clarity of low-resolution product images and optimizing visual presentation. It supports output in formats such as PNG, JPG, and BMP.
3655
+ * # 4. Effect comparison
3660
3656
  * | **Original image** | **Capabilities used** | **Output image** |
3661
3657
  * | --- | --- | --- |
3662
- * | ![image.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/3c1af966-8a5a-491d-9bfa-aefc321c9bed.png) | * Intelligent detection
3658
+ * | ![image.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/3c1af966-8a5a-491d-9bfa-aefc321c9bed.png) | * Intelligent recognition
3663
3659
  *
3664
- * * Image Translation Pro
3660
+ * * Image translation Pro
3665
3661
  *
3666
3662
  * * Intelligent cropping | ![R1.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/c2579e3e-389e-4717-84e3-834e60fea6b0.png) |
3667
- * | ![image.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/a92d14a9-ecf4-4e77-ad96-a2fcdd95dd1d.png) | * Intelligent matting
3663
+ * | ![image.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/a92d14a9-ecf4-4e77-ad96-a2fcdd95dd1d.png) | * Intelligent cutout
3668
3664
  *
3669
- * * Intelligent removal
3665
+ * * Intelligent erasure
3670
3666
  *
3671
3667
  * * Intelligent cropping | ![R1.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/f631f578-3496-4aa1-b5ef-cd1ea603e049.png) |
3672
- * | ![C2.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/12965cf8-cfdf-4a85-95fe-fdfcefcaa4ae.png) | * Intelligent detection
3668
+ * | ![C2.png](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/12965cf8-cfdf-4a85-95fe-fdfcefcaa4ae.png) | * Intelligent recognition
3673
3669
  *
3674
- * * Intelligent removal
3670
+ * * Intelligent erasure
3675
3671
  *
3676
- * * HD upscaling | ![R2.jpeg](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/95d29e79-8622-45fb-b284-cd580fe40ea0.jpeg) |
3672
+ * * High-definition upscaling | ![R2.jpeg](https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/NpQlK5jRe9w5jqDv/img/95d29e79-8622-45fb-b284-cd580fe40ea0.jpeg) |
3677
3673
  *
3678
3674
  * @param request - VisionFlowRequest
3679
3675
  * @returns VisionFlowResponse