Update pipeline/logic.py

pipeline/logic.py

@@ -29,32 +29,50 @@ def detect_and_crop(image: Image.Image, object_name: str, models: dict) -> Image
     image_np = np.array(image.convert("RGB"))
     height, width = image_np.shape[:2]
     prompt = [[f"a {object_name}"]]
-    inputs = models['processor_gnd'](images=image, text=prompt, return_tensors="pt").to(models['device'])
+
+    inputs = models['processor_gnd'](
+        images=image,
+        text=prompt,
+        return_tensors="pt"
+    ).to(models['device'])
+
     with torch.no_grad():
         outputs = models['model_gnd'](**inputs)
+
+    # Updated signature: use threshold and text_threshold, no box_threshold
     results = models['processor_gnd'].post_process_grounded_object_detection(
-        outputs, inputs.input_ids, box_threshold=0.4, text_threshold=0.3, target_sizes=[(height, width)]
+        outputs=outputs,
+        input_ids=inputs.input_ids,
+        threshold=0.4,
+        text_threshold=0.3,
+        target_sizes=[(height, width)]
     )
+
     if not results or len(results[0]['boxes']) == 0:
         print(" [Detect & Crop] ⚠ Warning: Grounding DINO did not detect the object. Using full image.")
         return image
+
     result = results[0]
     scores = result['scores']
     max_idx = int(torch.argmax(scores))
     box = result['boxes'][max_idx].cpu().numpy().astype(int)
     print(f" [Detect & Crop] ✅ Object detected with confidence: {scores[max_idx]:.2f}, Box: {box}")
+
     x1, y1, x2, y2 = box
     models['predictor'].set_image(image_np)
     box_prompt = np.array([[x1, y1, x2, y2]])
     masks, _, _ = models['predictor'].predict(box=box_prompt, multimask_output=False)
     mask = masks[0]
     mask_bool = mask > 0
+
     cropped_img_rgba = np.zeros((height, width, 4), dtype=np.uint8)
     cropped_img_rgba[:, :, :3] = image_np
     cropped_img_rgba[:, :, 3] = mask_bool * 255
     cropped_img_rgba = cropped_img_rgba[y1:y2, x1:x2]
+
     return Image.fromarray(cropped_img_rgba, 'RGBA')
 
+
 def extract_features(segmented_image: Image.Image) -> dict:
     image_rgba = np.array(segmented_image)
     if image_rgba.shape[2] != 4: raise ValueError("Segmented image must be RGBA")