nexaai 1.0.22__cp310-cp310-macosx_14_0_universal2.whl → 1.0.23__cp310-cp310-macosx_14_0_universal2.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=gOd7sNsqEESopw_24xgnOSkIRENrk4Fa-RMtmVv62eA,2421
-nexaai/_stub.cpython-310-darwin.so,sha256=OZuKXZRiNGMjNupMnGLv14pYMIoOzTxHGHflS4zUk2I,66768
-nexaai/_version.py,sha256=fWzVuSja_lnFQfiXurvM3Nq0-fbC2d6ueAZkQQ-1E6Q,139
+nexaai/_stub.cpython-310-darwin.so,sha256=8Qjo0gLyVm2O7jvohJd_JP0-7ScQif6dMrm6efSCKJw,66768
+nexaai/_version.py,sha256=j_sMUOs-rlDDzHhGRBuKZSz_Ij3I4oXriuZWuCat46U,139
 nexaai/asr.py,sha256=wqtq71cxIMGE4KvOIYZebHdWik8dy4LyKrDI98PDvzQ,2294
 nexaai/base.py,sha256=N8PRgDFA-XPku2vWnQIofQ7ipz3pPlO6f8YZGnuhquE,982
 nexaai/common.py,sha256=MRWZ6a7pnci_OUHxZRm3YqgKLAtZFD7b88STYDfeIF8,3460
-nexaai/cv.py,sha256=gpE3F__6bjh8OQKNJZs-QrBuCxqMj2eH-u6HR90vGZE,3302
+nexaai/cv.py,sha256=WbyFzvByx6hOoR2ZrGSm6i4uG0ubxqeMZ9x71kG6Des,3338
 nexaai/embedder.py,sha256=lXOT16PEvd_hT23d77dZH38VHNOAk-3JvoOUdQTEaGI,2552
 nexaai/image_gen.py,sha256=MkGw1HXqqv8cJzbiGERNPKFXfq9vMOlvuq0pgekXw68,4385
 nexaai/llm.py,sha256=-agVJuj0FOaDvDiT-fFSOpoyVt-MpNudBucsod3Vp1M,3673
@@ -20,8 +20,9 @@ nexaai/asr_impl/pybind_asr_impl.py,sha256=FLOWIph37q_nIiNx8xYi-VnhQ6CrPuc4HFAJZQ
 nexaai/binds/__init__.py,sha256=2-Rr0NwyWygqwS8Xlxq0BJ2ltyID-WbGuzEYNlSanCI,155
 nexaai/binds/asr_bind.cpython-310-darwin.so,sha256=nr1BqGYRCooj2k5khaIvGG06_GrzudMXa9xa15UQ-88,200768
 nexaai/binds/common_bind.cpython-310-darwin.so,sha256=zxJuD0nSV--VZKxBfWZUavU7_bHj_JTi0FhkjvG4VJw,235264
+nexaai/binds/cv_bind.cpython-310-darwin.so,sha256=-6Nj7bm03pWVUNEa4jYQW7Y7F5LuFEmK9Dl5nK0GDEU,250640
 nexaai/binds/embedder_bind.cpython-310-darwin.so,sha256=nn3kINQUNyXIzmcU1olLg8RlkZYyIs_RtIlCk6OH1ds,202064
-nexaai/binds/libnexa_bridge.dylib,sha256=SOnKb1axm7bJtwLEqV3DAIN9z0KCZWzABC3X-IQis48,291224
+nexaai/binds/libnexa_bridge.dylib,sha256=uxNHsoxVd5Kdvn045unNpq8McqcYPb5btuQMfcrqTlc,291224
 nexaai/binds/llm_bind.cpython-310-darwin.so,sha256=YP_GzIPBb4OMNmI-VMYgorx5g31oorcdBI3XQpi3jKI,182784
 nexaai/binds/rerank_bind.cpython-310-darwin.so,sha256=GJmffOLo9A48S_pMG2CtHyhbamtt97QikSSCXk3LqiM,183920
 nexaai/binds/vlm_bind.cpython-310-darwin.so,sha256=JEoQbyOkMXVK4WyBCqKBHbyXHtTiTWT18UkZMFvMh4k,182704
@@ -195,14 +196,14 @@ nexaai/binds/nexaml/libmp3lame.0.dylib,sha256=G-21u3MAZ2hiv1fbPEpOUGdToZfLtz2XI6
 nexaai/binds/nexaml/libmpg123.0.dylib,sha256=L4AgLcdcjGOQkEovUGDCNlH959500i0GkKBhqiHxBY4,306464
 nexaai/binds/nexaml/libnexa-mm-process.dylib,sha256=WCEgYK13YBfg-DiaGXXJ00-QYo0ucEP-QUS1jVEUR_k,10222392
 nexaai/binds/nexaml/libnexa-sampling.dylib,sha256=mWJ3jsF4bp3RwWebgjd4PalZqtLiXGK4Y5J9IkYqgvk,7957800
-nexaai/binds/nexaml/libnexa_plugin.dylib,sha256=7HglcsNk_2Stbpw22TpvItmPQY3vbQMvu-JOlDWOkHA,267224
+nexaai/binds/nexaml/libnexa_plugin.dylib,sha256=Qkdjyz4hG40BIa8-pngm-m05EeOzrXwFvBKa_0U3EVE,267224
 nexaai/binds/nexaml/libnexaproc.dylib,sha256=KNAyTwv2Tk94VNbUNFldDbnhQmJwv2ykysfvW_267DY,897248
 nexaai/binds/nexaml/libomp.dylib,sha256=RcvfaCz0XuqTYtVPFYyUm2V6S9YgUQ0x5pHATr9FnDk,754912
 nexaai/binds/nexaml/libqwen3-vl.dylib,sha256=zTKPJqAzSsUt4b2DOwkgcjysDSa9tAYY0m2HaxrIlLo,770784
 nexaai/binds/nexaml/libqwen3vl-vision.dylib,sha256=t0QfNZG5gpvHPBqX-Tdm1qjaXHCVROTtzL5bW78v4-I,490344
 nexaai/cv_impl/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nexaai/cv_impl/mlx_cv_impl.py,sha256=gKECQOv8iaWwG3bl7xeqVy2NN_9K7tYerIFzfn4eLo4,3228
-nexaai/cv_impl/pybind_cv_impl.py,sha256=uSmwBste4cT7c8DQmXzRLmzwDf773PAbXNYWW1UzVls,1064
+nexaai/cv_impl/pybind_cv_impl.py,sha256=YVIgNZe8FaUs6eH4IfirqhAyMy9RXqZL0eqVjGwSfO0,4761
 nexaai/embedder_impl/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nexaai/embedder_impl/mlx_embedder_impl.py,sha256=pFPraUAjm9EVvVbwIp1cjbtXUysF5pqxEcK2CAFvcDw,4639
 nexaai/embedder_impl/pybind_embedder_impl.py,sha256=lFpf0wI2d7kfO2GUyUuUS1U2L_PyZMJVGmAvF8EuQ0g,3653
@@ -569,7 +570,7 @@ nexaai/utils/quantization_utils.py,sha256=FYcNSAKGlBqFDUTx3jSKOr2lnq4nyiyC0ZG8oS
 nexaai/vlm_impl/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nexaai/vlm_impl/mlx_vlm_impl.py,sha256=sgHqnX5OCSGLccCnTuRiktIbqThNn3AAIvYE2_Dy4TI,10833
 nexaai/vlm_impl/pybind_vlm_impl.py,sha256=stJKHdhYhBuWUQkky-nHgCv625qDB_1geI3v5BLNGpM,9765
-nexaai-1.0.22.dist-info/METADATA,sha256=dgjz8lyw-1DssxtFV5wnxMO5P5iPnLVbOElwYtgrzic,1184
-nexaai-1.0.22.dist-info/WHEEL,sha256=T2p57lol9__xkoU6aJTyN1Pm43ZpRU3q6km7mIbrAMs,114
-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=E_PNlK0fIcVhcTkwQRaaY6TZ36LeRR7bRxFiy8LV7-g,1184
+nexaai-1.0.23.dist-info/WHEEL,sha256=T2p57lol9__xkoU6aJTyN1Pm43ZpRU3q6km7mIbrAMs,114
+nexaai-1.0.23.dist-info/top_level.txt,sha256=LRE2YERlrZk2vfuygnSzsEeqSknnZbz3Z1MHyNmBU4w,7
+nexaai-1.0.23.dist-info/RECORD,,