nexaai 1.0.22__cp310-cp310-win_amd64.whl → 1.0.23__cp310-cp310-win_amd64.whl

nexaai/_version.py

@@ -1,4 +1,4 @@
 # This file is generated by CMake from _version.py.in
 # Do not modify this file manually - it will be overwritten
 
-__version__ = "1.0.22" 
+__version__ = "1.0.23" 

nexaai/cv.py

@@ -10,7 +10,7 @@ from nexaai.common import PluginID
 class BoundingBox:
     """Generic bounding box structure."""
     x: float  # X coordinate (normalized or pixel, depends on model)
-    y: float  # Y coordinate (normalized or pixel, depends on model) 
+    y: float  # Y coordinate (normalized or pixel, depends on model)
     width: float  # Width
     height: float  # Height
 
@@ -24,7 +24,8 @@ class CVResult:
     confidence: float = 0.0  # Confidence score [0.0-1.0]
     bbox: Optional[BoundingBox] = None  # Bounding box (example: YOLO)
     text: Optional[str] = None  # Text result (example: OCR)
-    embedding: Optional[List[float]] = None  # Feature embedding (example: CLIP embedding)
+    # Feature embedding (example: CLIP embedding)
+    embedding: Optional[List[float]] = None
     embedding_dim: int = 0  # Embedding dimension
 
 
@@ -70,22 +71,23 @@ class CVModel(BaseModel):
 
     @classmethod
     def _load_from(cls,
-                   _: str,  # TODO: remove this argument, this is a hack to make api design happy
-                   config: CVModelConfig,
+                   local_path: str,
+                   model_name: Optional[str] = None,
+                   m_cfg: CVModelConfig = CVModelConfig(CVCapabilities.OCR),
                    plugin_id: Union[PluginID, str] = PluginID.LLAMA_CPP,
                    device_id: Optional[str] = None,
                    **kwargs
-        ) -> 'CVModel':
+                   ) -> 'CVModel':
         """Load CV model from configuration, routing to appropriate implementation."""
-        # Check plugin_id value for routing - handle both enum and string
-        plugin_value = plugin_id.value if isinstance(plugin_id, PluginID) else plugin_id
-        
+        plugin_value = plugin_id.value if isinstance(
+            plugin_id, PluginID) else plugin_id
+
         if plugin_value == "mlx":
             from nexaai.cv_impl.mlx_cv_impl import MLXCVImpl
-            return MLXCVImpl._load_from(config, plugin_id, device_id)
+            return MLXCVImpl._load_from(local_path, model_name, m_cfg, plugin_id, device_id, **kwargs)
         else:
             from nexaai.cv_impl.pybind_cv_impl import PyBindCVImpl
-            return PyBindCVImpl._load_from(config, plugin_id, device_id)
+            return PyBindCVImpl._load_from(local_path, model_name, m_cfg, plugin_id, device_id, **kwargs)
 
     @abstractmethod
     def infer(self, input_image_path: str) -> CVResults:

nexaai/cv_impl/pybind_cv_impl.py

@@ -1,32 +1,124 @@
-from typing import Optional, Union
+from typing import Optional, Union, List
+import os
 
-from nexaai.common import PluginID
-from nexaai.cv import CVModel, CVModelConfig, CVResults
+from nexaai.common import PluginID, ModelConfig
+from nexaai.cv import CVModel, CVModelConfig, CVResults, CVResult, BoundingBox, CVCapabilities
+from nexaai.binds import cv_bind, common_bind
+from nexaai.runtime import _ensure_runtime
 
 
 class PyBindCVImpl(CVModel):
-    def __init__(self):
-        """Initialize PyBind CV implementation."""
+    def __init__(self, handle: any, m_cfg: ModelConfig = ModelConfig()):
+        """Private constructor, should not be called directly."""
         super().__init__()
-        # TODO: Add PyBind-specific initialization
+        self._handle = handle  # This is a py::capsule
+        self._model_config = None
 
     @classmethod
     def _load_from(cls,
-                   config: CVModelConfig,
+                   local_path: str,  # This is the local path after auto_download_model processing
+                   model_name: Optional[str] = None,
+                   m_cfg: CVModelConfig = CVModelConfig(CVCapabilities.OCR),
                    plugin_id: Union[PluginID, str] = PluginID.LLAMA_CPP,
-                   device_id: Optional[str] = None
-        ) -> 'PyBindCVImpl':
+                   device_id: Optional[str] = None,
+                   **kwargs
+                   ) -> 'PyBindCVImpl':
         """Load CV model from configuration using PyBind backend."""
-        # TODO: Implement PyBind CV loading
-        instance = cls()
-        return instance
+        _ensure_runtime()
+
+        config = cv_bind.CVModelConfig()
+        config.capabilities = cv_bind.CVCapabilities(m_cfg.capabilities)
+        if m_cfg.det_model_path is not None:
+            config.det_model_path = m_cfg.det_model_path
+        else:
+            config.det_model_path = local_path
+
+        print("local_path: ", local_path)
+        print("m_cfg.rec_model_path: ", m_cfg.rec_model_path)
+        if m_cfg.rec_model_path is not None:
+            config.rec_model_path = m_cfg.rec_model_path
+        else:
+            config.rec_model_path = local_path
+        print("config.rec_model_path: ", config.rec_model_path)
+        
+        if m_cfg.char_dict_path is not None:
+            config.char_dict_path = m_cfg.char_dict_path
+
+        if m_cfg.model_path is not None:
+            config.model_path = m_cfg.model_path
+
+        if m_cfg.system_library_path is not None:
+            config.system_library_path = m_cfg.system_library_path
+
+        plugin_id_str = plugin_id.value if isinstance(
+            plugin_id, PluginID) else str(plugin_id)
+
+        model_name_to_use = model_name if model_name else local_path
+        handle = cv_bind.ml_cv_create(
+            model_name=model_name_to_use,
+            config=config,
+            plugin_id=plugin_id_str,
+            device_id=device_id,
+            license_id=None,
+            license_key=None
+        )
+
+        return cls(handle, m_cfg)
 
     def eject(self):
-        """Destroy the model and free resources."""
-        # TODO: Implement PyBind CV cleanup
-        pass
+        """Release the model from memory."""
+        # py::capsule handles cleanup automatically
+        if hasattr(self, '_handle') and self._handle is not None:
+            del self._handle
+            self._handle = None
 
     def infer(self, input_image_path: str) -> CVResults:
         """Perform inference on image."""
-        # TODO: Implement PyBind CV inference
-        raise NotImplementedError("PyBind CV inference not yet implemented")
+        if self._handle is None:
+            raise RuntimeError("CV model not loaded. Call _load_from first.")
+
+        if not os.path.exists(input_image_path):
+            raise FileNotFoundError(
+                f"Input image not found: {input_image_path}")
+
+        try:
+            # Perform inference using the binding
+            result_dict = cv_bind.ml_cv_infer(
+                handle=self._handle,
+                input_image_path=input_image_path
+            )
+
+            # Convert result dictionary to CVResults
+            results = []
+            for result_data in result_dict["results"]:
+                # Create bounding box if present
+                bbox = None
+                if "bbox" in result_data and result_data["bbox"] is not None:
+                    bbox_data = result_data["bbox"]
+                    bbox = BoundingBox(
+                        x=bbox_data["x"],
+                        y=bbox_data["y"],
+                        width=bbox_data["width"],
+                        height=bbox_data["height"]
+                    )
+
+                # Create CV result
+                cv_result = CVResult(
+                    image_paths=result_data.get("image_paths"),
+                    image_count=result_data.get("image_count", 0),
+                    class_id=result_data.get("class_id", 0),
+                    confidence=result_data.get("confidence", 0.0),
+                    bbox=bbox,
+                    text=result_data.get("text"),
+                    embedding=result_data.get("embedding"),
+                    embedding_dim=result_data.get("embedding_dim", 0)
+                )
+                results.append(cv_result)
+
+            return CVResults(
+                results=results,
+                result_count=result_dict["result_count"]
+            )
+
+        except Exception as e:
+            raise RuntimeError(f"CV inference failed: {str(e)}")

{nexaai-1.0.22.dist-info → nexaai-1.0.23.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nexaai
-Version: 1.0.22
+Version: 1.0.23
 Summary: Python bindings for NexaSDK C-lib backend
 Author-email: "Nexa AI, Inc." <dev@nexa.ai>
 Project-URL: Homepage, https://github.com/NexaAI/nexasdk-bridge

{nexaai-1.0.22.dist-info → nexaai-1.0.23.dist-info}/RECORD

@@ -1,10 +1,10 @@
 nexaai/__init__.py,sha256=Yt2YPVcRfwdA6DQoUR4ZI34CHBUYoY2cVsrG7Zp6IrI,2516
-nexaai/_stub.cp310-win_amd64.pyd,sha256=KEyu4whFbX7ahYkHxUZE9Bvb68yBBNWGR0CNQ-4fCvs,10752
-nexaai/_version.py,sha256=ivY4bEQ0jgel7C15vIMosfWJgA8mHuZ-du5RS_0fPko,143
+nexaai/_stub.cp310-win_amd64.pyd,sha256=krxve5v7zNDe00GvywzSZC7LjdPJqPd6jm75Zd7sobE,10752
+nexaai/_version.py,sha256=jvuw8FsRzBuy2gWtiNcMHIovBrvEB6BUEhxloSiR5bE,143
 nexaai/asr.py,sha256=p_bTRqESa-4g6V3HGFzi8vqn8H4Q6yTLiLjL0yqVr3I,2362
 nexaai/base.py,sha256=qQBCiQVNzgpkQjZX9aiFDEdbAAe56TROKC3WnWra2Zg,1021
 nexaai/common.py,sha256=XRwjvRNxS2xRVb0CMR5NbPQ4E-PAewJ0m6VGbZOw19A,3565
-nexaai/cv.py,sha256=PhgqK5hO8FlXLR6QxE_pHtXjdtzLSEPzkCda-qrNpOs,3395
+nexaai/cv.py,sha256=TI6VlaztYWpvgkF7tLvQWYgCaNgGAfXJXsbCLavM6wE,3433
 nexaai/embedder.py,sha256=fzIdlE52SSVNurIFZgf9gKtRf2SXysARsETQmkiCWs0,2625
 nexaai/image_gen.py,sha256=_kLLCbvhc30a-YGidkB13ZXYoPOS-sKMo8bKVRkUado,4526
 nexaai/llm.py,sha256=woB98NQmaNf3_BH0Ogig3niU1rcD8YtHPiSpzyrbVmo,3771
@@ -18,37 +18,38 @@ nexaai/asr_impl/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nexaai/asr_impl/mlx_asr_impl.py,sha256=XwMX3LYMeulp8cDS0TCCYcjvttFHAyDWQ_oMvABwQmI,3349
 nexaai/asr_impl/pybind_asr_impl.py,sha256=I-J_pp4Tdafa-ipp34q2dN4dbL905nlpBYVBGLQwLVI,4807
 nexaai/binds/__init__.py,sha256=cDlzRJF5pq0YmiwmUvdNO9CrvYrBdkYWR4xXrEbwLyU,161
-nexaai/binds/asr_bind.cp310-win_amd64.pyd,sha256=m50-FsvEkrfVFtf6Rk1AVPppkCrELc71Smc2W6ebs2M,192000
-nexaai/binds/common_bind.cp310-win_amd64.pyd,sha256=hy-gEksGbyuSa44XGe5gmfwvmfm0iZJuTNOd_v0Xo4A,205824
-nexaai/binds/embedder_bind.cp310-win_amd64.pyd,sha256=wo0-I8V9uUBCW_9Mof_42vd-QK4sTUPfES3bg5AIlFg,183808
+nexaai/binds/asr_bind.cp310-win_amd64.pyd,sha256=vl2y1zBTCoX1gD9yIMYs0B3iuNSLR1RhA4o56umN8TA,192000
+nexaai/binds/common_bind.cp310-win_amd64.pyd,sha256=5tmncxGObCeb7WX37akWRyubL2fCtbNG3u8S0aLZyWs,205824
+nexaai/binds/cv_bind.cp310-win_amd64.pyd,sha256=BVdIMSx64jl-f4u8a2ueiF23upTVsTPpQPpzuZtLoZY,230912
+nexaai/binds/embedder_bind.cp310-win_amd64.pyd,sha256=Y1jBOz9nezR1MHjYIGGZOMyvSznJkyAhgQJJ7Rx3bNw,183808
 nexaai/binds/libcrypto-3-x64.dll,sha256=X7hDuQPMn6eRe58dTGJxTmRPKFWgd5i6Gjoy9QPLmPg,7315968
 nexaai/binds/libssl-3-x64.dll,sha256=GRw0cOaGUaR2QHRvBFuU_9S9iNgkyVgm0NIpZMnReAw,1313792
-nexaai/binds/llm_bind.cp310-win_amd64.pyd,sha256=KdcZF5shnqqfqogeiEJ-11OUe1os7rr22bW25Xbv_I4,166400
-nexaai/binds/nexa_bridge.dll,sha256=VYTa4zclMKJC2AfL155QLxJzH4wcO_8O5O7Rd8Dj5LQ,190976
-nexaai/binds/rerank_bind.cp310-win_amd64.pyd,sha256=BYf5QaRA8-Wjar7FuE890NF6LGkage6FZRaUPLCt3B8,178176
-nexaai/binds/vlm_bind.cp310-win_amd64.pyd,sha256=XyBTpjgYMfoD10YuRnfB3Fgivzs7RSWSD9yR6eN2k6o,171520
-nexaai/binds/cpu_gpu/ggml-base.dll,sha256=vcESAaud2k5HfVS4xjdUHlb8hPDzw7qVM7huYykDGws,532480
-nexaai/binds/cpu_gpu/ggml-cpu.dll,sha256=zbXppL8mtjbVGxFVV7hIDXb4Vi7oqQi-kndzb8mp3kI,672768
-nexaai/binds/cpu_gpu/ggml-cuda.dll,sha256=R66N-OKwvooY0jhTfhxEBKF5-vfraG5RC6ABVXJQ-e4,313528832
-nexaai/binds/cpu_gpu/ggml-vulkan.dll,sha256=hOEeLexuKD1uenpijMqTOe4AmYSMGe_jg6JqagOlAAE,36627968
-nexaai/binds/cpu_gpu/ggml.dll,sha256=Qlypa-cCdBcnXkgg2ErVRo7OI4Vsfsg-r1E8oeke4sk,66560
-nexaai/binds/cpu_gpu/mtmd.dll,sha256=yP6ze4ks_j8Na55RKWYjigybP5sMchNLIA_cgQvR_Zw,561152
-nexaai/binds/cpu_gpu/nexa_cpu_gpu.dll,sha256=0a2Yf-NqIuU3wDeQHiirTqIaXwmqnWIJ3RCfQh7bsUM,1611776
-nexaai/binds/cpu_gpu/nexa_plugin.dll,sha256=tDt7ys1fXVhno3VA2Z62HL79oyv087vmH3QUE0eJMUk,1424384
-nexaai/binds/nexaml/ggml-base.dll,sha256=vcESAaud2k5HfVS4xjdUHlb8hPDzw7qVM7huYykDGws,532480
-nexaai/binds/nexaml/ggml-cpu.dll,sha256=zbXppL8mtjbVGxFVV7hIDXb4Vi7oqQi-kndzb8mp3kI,672768
-nexaai/binds/nexaml/ggml-cuda.dll,sha256=R66N-OKwvooY0jhTfhxEBKF5-vfraG5RC6ABVXJQ-e4,313528832
-nexaai/binds/nexaml/ggml-vulkan.dll,sha256=hOEeLexuKD1uenpijMqTOe4AmYSMGe_jg6JqagOlAAE,36627968
-nexaai/binds/nexaml/ggml.dll,sha256=Qlypa-cCdBcnXkgg2ErVRo7OI4Vsfsg-r1E8oeke4sk,66560
+nexaai/binds/llm_bind.cp310-win_amd64.pyd,sha256=GVzooNvq_hGX8sgZ0OzlrIji92kQsNEy0DrRETP-Xi0,166400
+nexaai/binds/nexa_bridge.dll,sha256=ddpimInP4T8NZt377hMCjhLgV7EBPVlSEpdldXSVqpo,190976
+nexaai/binds/rerank_bind.cp310-win_amd64.pyd,sha256=hQY6MEabDIg029J_Ut2H5jCL8ktACb8X8m8tQlmL9Jk,178176
+nexaai/binds/vlm_bind.cp310-win_amd64.pyd,sha256=VdUSKpjBe-PA2Yp30-tbxkKr1DLF_wFE2CvZYyMnWHY,171520
+nexaai/binds/cpu_gpu/ggml-base.dll,sha256=KrfapxFjVod-m3zkbvplZgu79nGICpB8u_TYv_JpuUE,532480
+nexaai/binds/cpu_gpu/ggml-cpu.dll,sha256=ZnIGD3CZNDAEuXhm_OEpZWMmMdeM0PqdGFdnIPuAMeI,672768
+nexaai/binds/cpu_gpu/ggml-cuda.dll,sha256=erOwNYLmLg42-0m7ZjwI_-6gUbx7_6tsMq8nUljgU78,313528832
+nexaai/binds/cpu_gpu/ggml-vulkan.dll,sha256=lUYTXN9hXt8NVz9QieCAcDKPt2NvokWlFYQrO-xbHzo,36627968
+nexaai/binds/cpu_gpu/ggml.dll,sha256=wROpeY7dNTg3grpUscFhPoByniT5NXk8hNMZl6DfSFc,66560
+nexaai/binds/cpu_gpu/mtmd.dll,sha256=3DTtZOm_ItGjHm8RHsSqI4t6_rXQqeBkzSs0vetZ3GQ,561152
+nexaai/binds/cpu_gpu/nexa_cpu_gpu.dll,sha256=pm0GwgepM78cYM9Ae5Sn5Uygeh_li-7hLhCd2Rx-MZQ,1611776
+nexaai/binds/cpu_gpu/nexa_plugin.dll,sha256=pdMax4xDZSNqGTqdDhugTbHwggj_OMD7aB-YanjCQH4,1424384
+nexaai/binds/nexaml/ggml-base.dll,sha256=KrfapxFjVod-m3zkbvplZgu79nGICpB8u_TYv_JpuUE,532480
+nexaai/binds/nexaml/ggml-cpu.dll,sha256=ZnIGD3CZNDAEuXhm_OEpZWMmMdeM0PqdGFdnIPuAMeI,672768
+nexaai/binds/nexaml/ggml-cuda.dll,sha256=erOwNYLmLg42-0m7ZjwI_-6gUbx7_6tsMq8nUljgU78,313528832
+nexaai/binds/nexaml/ggml-vulkan.dll,sha256=lUYTXN9hXt8NVz9QieCAcDKPt2NvokWlFYQrO-xbHzo,36627968
+nexaai/binds/nexaml/ggml.dll,sha256=wROpeY7dNTg3grpUscFhPoByniT5NXk8hNMZl6DfSFc,66560
 nexaai/binds/nexaml/nexa-mm-process.dll,sha256=DDD5qPbelZAIrRWGggOd3Pg9U_fh_ZEYIplwqiyD5LY,4643328
 nexaai/binds/nexaml/nexa-sampling.dll,sha256=Notkz287laSUG2_ED3oFXWVLJM3t7x_USkvX6wKVOLA,4265984
-nexaai/binds/nexaml/nexa_plugin.dll,sha256=dNr9-h_2jCAmVm74x0gT7JjFEuEv1MiSpZXevjUS-zo,604160
+nexaai/binds/nexaml/nexa_plugin.dll,sha256=7ZJYrF1uCgnNnIbXSU0fSiyUlau_MpgSPFKjR2wPjzs,604160
 nexaai/binds/nexaml/nexaproc.dll,sha256=rRcdSfhkxnRwktXW4lp7Np8sNY_uefNXYwAmPp34AHg,2670080
 nexaai/binds/nexaml/qwen3-vl.dll,sha256=k0wPZNR6VejeR32KuFgCVyj7t7YaEmZOKhj9lwMw0-s,5874176
 nexaai/binds/nexaml/qwen3vl-vision.dll,sha256=5AX4MtAOWTcXSAkS2qqvl2pFti4HR55dtkCaOO03Evc,1063424
 nexaai/cv_impl/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nexaai/cv_impl/mlx_cv_impl.py,sha256=QLd_8w90gtxH8kmssaDYatCTRvQNIJuUGKZNnYrmx6E,3317
-nexaai/cv_impl/pybind_cv_impl.py,sha256=aSOCAxmHrwJbEkSN6VX3Cykqlj_9RIpVrZXILul04GA,1096
+nexaai/cv_impl/pybind_cv_impl.py,sha256=8jGGHPWF0-z8ch2BPn7Vn2pIjzUJ41b208Op17ifiYs,4885
 nexaai/embedder_impl/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nexaai/embedder_impl/mlx_embedder_impl.py,sha256=WWTQrUWEMq1kLGg4CBQyu59zD2hIptghVEQXQAcx7hM,4757
 nexaai/embedder_impl/pybind_embedder_impl.py,sha256=Xsf4nczEzpRDqdx96DEu5GwLxGcSV06ZVnEV3a5-q68,3749
@@ -73,7 +74,7 @@ nexaai/utils/quantization_utils.py,sha256=FxnZ6-uAE_bl_vQ5jsRXlpU0NBn-U4Y8iN9_O6
 nexaai/vlm_impl/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nexaai/vlm_impl/mlx_vlm_impl.py,sha256=gI09QJcok68QG-36csXr3c2yPCZXTE0dlq1MUBEo_Ms,11091
 nexaai/vlm_impl/pybind_vlm_impl.py,sha256=w8gAZNW2lk3mCKolg4iKgD58RhHCEk4v9aYI1RwVGqc,10021
-nexaai-1.0.22.dist-info/METADATA,sha256=1THXwOXkS8oTllWqv2QjVjcJ8J9VzB_MSCmB0W8uWjo,1215
-nexaai-1.0.22.dist-info/WHEEL,sha256=KUuBC6lxAbHCKilKua8R9W_TM71_-9Sg5uEP3uDWcoU,101
-nexaai-1.0.22.dist-info/top_level.txt,sha256=LRE2YERlrZk2vfuygnSzsEeqSknnZbz3Z1MHyNmBU4w,7
-nexaai-1.0.22.dist-info/RECORD,,
+nexaai-1.0.23.dist-info/METADATA,sha256=ghgqII8_rVkGTKEww-GkKbxEtYSuGl5oWOlqoQXKqBQ,1215
+nexaai-1.0.23.dist-info/WHEEL,sha256=KUuBC6lxAbHCKilKua8R9W_TM71_-9Sg5uEP3uDWcoU,101
+nexaai-1.0.23.dist-info/top_level.txt,sha256=LRE2YERlrZk2vfuygnSzsEeqSknnZbz3Z1MHyNmBU4w,7
+nexaai-1.0.23.dist-info/RECORD,,