Update app.py

app.py

@@ -62,8 +62,8 @@ def compute_iou(boxA, boxB):
         return 0
     return interArea / float(boxAArea + boxBArea - interArea)
 
-# Lowering the NMS threshold to 0.5 (from 0.7) to allow more distinct boxes when they are adjacent.
-def custom_nms(preds, iou_threshold=0.5):
+# Lower the NMS threshold to 0.3 so that adjacent boxes are less likely to be merged.
+def custom_nms(preds, iou_threshold=0.3):
     preds = sorted(preds, key=lambda x: x["confidence"], reverse=True)
     filtered_preds = []
     for pred in preds:
@@ -76,8 +76,10 @@ def custom_nms(preds, iou_threshold=0.5):
             filtered_preds.append(pred)
     return filtered_preds
 
-# The process_image function merges robust box detection (with updated prediction parameters)
-# and ArUco marker detection with refined conversion factor computation.
+# The process_image function now uses:
+#   - Roboflow prediction parameters: confidence=50 and a lower overlap=10.
+#   - A custom NMS with IoU threshold of 0.3.
+#   - ArUco marker detection for conversion factor computation.
 def process_image(job_id, image_path, object_type, multiplier):
     try:
         jobs[job_id]['progress'] = 10
@@ -101,19 +103,19 @@ def process_image(job_id, image_path, object_type, multiplier):
                 return
 
             # --- BOX DETECTION ---
-            # Upscale if image dimensions are small.
+            # Upscale if the image is small.
             scale_factor = 1
             if img_width < 1000 or img_height < 1000:
                 scale_factor = 2
 
-            # Use improved parameters: confidence=50 and overlap=20.
+            # Use improved parameters: confidence=50 and overlap=10 (lowered overlap).
             if scale_factor > 1:
                 upscaled_image = cv2.resize(image, None, fx=scale_factor, fy=scale_factor, interpolation=cv2.INTER_LINEAR)
                 temp_path = "upscaled.jpg"
                 cv2.imwrite(temp_path, upscaled_image)
-                results = box_model.predict(temp_path, confidence=50, overlap=20).json()
+                results = box_model.predict(temp_path, confidence=50, overlap=10).json()
             else:
-                results = box_model.predict(image_path, confidence=50, overlap=20).json()
+                results = box_model.predict(image_path, confidence=50, overlap=10).json()
 
             predictions = results.get("predictions", [])
             processed_preds = []
@@ -130,12 +132,12 @@ def process_image(job_id, image_path, object_type, multiplier):
                         width = prediction["width"]
                         height = prediction["height"]
 
-                    # Convert from center coordinates to corner coordinates.
+                    # Convert center-based coordinates to corner-based bounding box.
                     x1 = int(round(x - width / 2))
                     y1 = int(round(y - height / 2))
                     x2 = int(round(x + width / 2))
                     y2 = int(round(y + height / 2))
-                    # Clamp to image dimensions.
+                    # Clamp coordinates within the image.
                     x1 = max(0, min(x1, img_width - 1))
                     y1 = max(0, min(y1, img_height - 1))
                     x2 = max(0, min(x2, img_width - 1))
@@ -148,8 +150,8 @@ def process_image(job_id, image_path, object_type, multiplier):
                 except Exception as e:
                     continue
 
-            # Apply NMS with a lower IoU threshold (0.5) to separate adjacent boxes.
-            box_detections = custom_nms(processed_preds, iou_threshold=0.5)
+            # Apply custom NMS with an IoU threshold of 0.3.
+            box_detections = custom_nms(processed_preds, iou_threshold=0.3)
             jobs[job_id]['progress'] = 60
 
             # --- ARUCO MARKER DETECTION & SIZE CONVERSION ---
@@ -173,7 +175,7 @@ def process_image(job_id, image_path, object_type, multiplier):
                     width_pixels = max_x - min_x
                     height_pixels = max_y - min_y
                     if width_pixels > 0 and height_pixels > 0:
-                        # Use the average of the width and height conversion factors.
+                        # Use the average conversion factor from width and height.
                         conversion_factor = (marker_real_width_cm / width_pixels + marker_real_width_cm / height_pixels) / 2
                     else:
                         conversion_factor = None
@@ -244,7 +246,7 @@ def process_image(job_id, image_path, object_type, multiplier):
                 jobs[job_id]['result'] = {"error": "Error during YOLOv5 inference."}
                 return
 
-        # Build summary text drawn on the image
+        # Draw summary text on the image
         detection_counts = Counter(det["class"] for det in detection_info)
         if detection_counts:
             top_text = ", ".join(f"{cls}: {count}" for cls, count in detection_counts.items())