show only masks > 1%

app.py

@@ -3,15 +3,11 @@ import spaces
 
 from pathlib import Path
 
-# import os
-# os.system("pip install .")
-
 import numpy as np
 import yaml
 from demo.demo_utils import DotDict, concat_instances, filter_instances, pose_nms, visualize_demo
 from demo.mm_utils import run_MMDetector, run_MMPose
 from mmdet.apis import init_detector
-from mmengine.logging import print_log
 from demo.sam2_utils import prepare_model as prepare_sam2_model
 from demo.sam2_utils import process_image_with_SAM
 
@@ -96,7 +92,6 @@ def process_image_with_BMP(
     rtmdet_result = None
     all_detections = None
     for iteration in range(bmp_config.num_bmp_iters):
-        print_log("BMP Iteration {}/{} started".format(iteration + 1, bmp_config.num_bmp_iters), logger="current")
 
         # Step 1: Detection
         det_instances = run_MMDetector(
@@ -106,9 +101,7 @@ def process_image_with_BMP(
             bbox_thr=DEFAULT_BBOX_THR,
             nms_thr=DEFAULT_NMS_THR,
         )
-        print_log("Detected {} instances".format(len(det_instances.bboxes)), logger="current")
         if len(det_instances.bboxes) == 0:
-            print_log("No detections found, skipping.", logger="current")
             continue
 
         # Step 2: Pose estimation
@@ -149,7 +142,6 @@ def process_image_with_BMP(
         keep_new_indices = [i - num_old_detections for i in keep_indices if i >= num_old_detections]
         keep_old_indices = [i for i in keep_indices if i < num_old_detections]
         if len(keep_new_indices) == 0:
-            print_log("No new instances passed pose NMS, skipping SAM refinement.", logger="current")
             continue
         # filter new detections and compute scores
         new_dets = filter_instances(pose_instances, keep_new_indices)
@@ -157,12 +149,6 @@ def process_image_with_BMP(
         old_dets = None
         if len(keep_old_indices) > 0:
             old_dets = filter_instances(all_detections, keep_old_indices)
-        print_log(
-            "Pose NMS reduced instances to {:d} ({:d}+{:d}) instances".format(
-                len(new_dets.bboxes) + num_old_detections, num_old_detections, len(new_dets.bboxes)
-            ),
-            logger="current",
-        )
 
         new_detections = process_image_with_SAM(
             DotDict(bmp_config.sam2.prompting),
@@ -186,8 +172,6 @@ def process_image_with_BMP(
 
         if iteration == 0:
             rtmdet_result = rtmdet_r
-        print_log("Iteration {} completed".format(iteration + 1), logger="current")
-
 
     _, _, bmp_result = visualize_demo(
         img.copy(),

demo/demo_utils.py

@@ -145,23 +145,23 @@ def _visualize_predictions(
     """
     vis_types = vis_type.split("+")
 
-    # # Filter-out small detections to make the image more clear
-    # new_bboxes = []
-    # new_scores = []
-    # new_masks = []
-    # new_poses = []
-    # size_thr = img.shape[0] * img.shape[1] * 0.01
-    # for bbox, score, mask, pose in zip(bboxes, scores, masks, poses):
-    #     area = mask.sum() # Assume binary mask. OK for demo purposes
-    #     if area > size_thr:
-    #         new_bboxes.append(bbox)
-    #         new_scores.append(score)
-    #         new_masks.append(mask)
-    #         new_poses.append(pose)
-    # bboxes = np.array(new_bboxes)
-    # scores = np.array(new_scores)
-    # masks = new_masks
-    # poses = new_poses
+    # Filter-out small detections to make the visualization more clear
+    new_bboxes = []
+    new_scores = []
+    new_masks = []
+    new_poses = []
+    size_thr = img.shape[0] * img.shape[1] * 0.01
+    for bbox, score, mask, pose in zip(bboxes, scores, masks, poses):
+        area = mask.sum() # Assume binary mask. OK for demo purposes
+        if area > size_thr:
+            new_bboxes.append(bbox)
+            new_scores.append(score)
+            new_masks.append(mask)
+            new_poses.append(pose)
+    bboxes = np.array(new_bboxes)
+    scores = np.array(new_scores)
+    masks = new_masks
+    poses = new_poses
     
     if mask_is_binary:
         poly_masks: List[Optional[List[np.ndarray]]] = []

mmpose/datasets/transforms/common_transforms.py

@@ -291,10 +291,8 @@ class MaskBackground(BaseTransform):
 
         if mask is None and not self.continue_on_failure:
             raise ValueError('No mask found in the results and self.continue_on_failure is set to False.')
-        print("\n\nI'm inside the MaskBackground transform")
 
         if mask is not None and self._do_masking():
-            print("MaskBackground transform will take place now\n\n")
             # Convert mask from polygons to binary mask
             try:            
                 mask_rle = Mask.frPyObjects(mask, results['img_shape'][0], results['img_shape'][1])