Upload export_medsiglip.py with huggingface_hub

export_medsiglip.py

@@ -10,10 +10,10 @@ Export trained MedSigLIP model to ONNX/CoreML/TFLite for mobile deployment
 #   "onnx>=1.14.0",
 #   "onnxscript>=0.1.0",
 #   "onnxruntime>=1.16.0",
-#   "onnxruntime-gpu>=1.16.0",
 #   "huggingface_hub>=0.20.0",
 #   "numpy>=1.24.0",
 #   "pillow>=10.0.0",
+#   "coremltools>=7.0",
 # ]
 # ///
 
@@ -281,9 +281,10 @@ def export_to_onnx(model, config: Config, quantize: bool = True):
 
 
 def export_to_coreml(model, config: Config):
-    """Export model to CoreML format for iOS"""
+    """Export model to CoreML format for iOS with Neural Engine optimization"""
     try:
         import coremltools as ct
+        from coremltools.models.neural_network import quantization_utils
     except ImportError:
         print("coremltools not installed. Skipping CoreML export.")
         print("Install with: pip install coremltools")
@@ -292,46 +293,83 @@ def export_to_coreml(model, config: Config):
     config.output_dir.mkdir(parents=True, exist_ok=True)
 
     model.eval()
+    model = model.cpu()  # Move to CPU for export
     wrapper = LaborViewExportWrapper(model)
     wrapper.eval()
 
     # Trace the model
+    print("Tracing model for CoreML...")
     dummy_input = torch.randn(1, 3, config.image_size, config.image_size)
-    traced_model = torch.jit.trace(wrapper, dummy_input)
+
+    with torch.no_grad():
+        traced_model = torch.jit.trace(wrapper, dummy_input)
 
     coreml_path = config.output_dir / "laborview_medsiglip.mlpackage"
+    coreml_fp16_path = config.output_dir / "laborview_medsiglip_fp16.mlpackage"
 
-    print(f"Exporting to CoreML: {coreml_path}")
+    print(f"Converting to CoreML: {coreml_path}")
 
-    # Convert to CoreML
-    mlmodel = ct.convert(
-        traced_model,
-        inputs=[
-            ct.TensorType(
-                name="pixel_values",
-                shape=(1, 3, config.image_size, config.image_size),
-                dtype=np.float32
-            )
-        ],
-        outputs=[
-            ct.TensorType(name="seg_probs"),
-            ct.TensorType(name="plane_pred")
-        ],
-        minimum_deployment_target=ct.target.iOS16,
-        compute_precision=ct.precision.FLOAT16  # Use FP16 for mobile
-    )
+    try:
+        # Convert to CoreML with FP32 first
+        mlmodel = ct.convert(
+            traced_model,
+            inputs=[
+                ct.TensorType(
+                    name="pixel_values",
+                    shape=(1, 3, config.image_size, config.image_size),
+                    dtype=np.float32
+                )
+            ],
+            outputs=[
+                ct.TensorType(name="seg_probs"),
+                ct.TensorType(name="plane_pred")
+            ],
+            minimum_deployment_target=ct.target.iOS16,
+            compute_units=ct.ComputeUnit.ALL  # Use Neural Engine + GPU + CPU
+        )
 
-    # Add metadata
-    mlmodel.author = "LaborView AI"
-    mlmodel.short_description = "Ultrasound segmentation for labor monitoring"
-    mlmodel.version = "1.0"
+        # Add metadata
+        mlmodel.author = "LaborView AI"
+        mlmodel.short_description = "MedSigLIP ultrasound segmentation for labor monitoring"
+        mlmodel.version = "1.0"
+
+        # Save FP32 version
+        mlmodel.save(str(coreml_path))
+        print(f"CoreML FP32 model saved: {coreml_path}")
+
+        # Create FP16 version for smaller size and faster inference
+        print("Creating FP16 quantized version...")
+        mlmodel_fp16 = ct.convert(
+            traced_model,
+            inputs=[
+                ct.TensorType(
+                    name="pixel_values",
+                    shape=(1, 3, config.image_size, config.image_size),
+                    dtype=np.float32
+                )
+            ],
+            outputs=[
+                ct.TensorType(name="seg_probs"),
+                ct.TensorType(name="plane_pred")
+            ],
+            minimum_deployment_target=ct.target.iOS16,
+            compute_units=ct.ComputeUnit.ALL,
+            compute_precision=ct.precision.FLOAT16
+        )
 
-    # Save
-    mlmodel.save(str(coreml_path))
+        mlmodel_fp16.author = "LaborView AI"
+        mlmodel_fp16.short_description = "MedSigLIP ultrasound segmentation (FP16)"
+        mlmodel_fp16.version = "1.0"
+        mlmodel_fp16.save(str(coreml_fp16_path))
+        print(f"CoreML FP16 model saved: {coreml_fp16_path}")
 
-    print(f"CoreML model saved!")
+        return coreml_path
 
-    return coreml_path
+    except Exception as e:
+        print(f"CoreML export failed: {e}")
+        import traceback
+        traceback.print_exc()
+        return None
 
 
 def verify_onnx_model(onnx_path: Path, config: Config):
@@ -378,7 +416,7 @@ def main():
     parser = argparse.ArgumentParser(description="Export MedSigLIP for edge deployment")
     parser.add_argument("--output-dir", type=str, default="./exports", help="Output directory")
     parser.add_argument("--quantize", action="store_true", default=True, help="Quantize to INT8")
-    parser.add_argument("--coreml", action="store_true", help="Export to CoreML")
+    parser.add_argument("--coreml", action="store_true", default=True, help="Export to CoreML")
     parser.add_argument("--verify", action="store_true", default=True, help="Verify exported model")
     args = parser.parse_args()