Upload 2 files

backend.py

@@ -141,23 +141,8 @@ def process_input(input_video_path):
         
         print(f"1. Extracting keypoints from video... [{time.time() - start_time:.2f} s since start]")
         video_frames, fps = load_video_cv(input_video_path)
-        
-        # Normalize high frame rate (e.g. 60 or 120 FPS from webcams/phones) to 30 FPS
-        if fps and fps > 32:
-            step = fps / 30.0
-            num_target_frames = int(len(video_frames) / step)
-            if num_target_frames > 0:
-                video_frames = [video_frames[int(i * step)] for i in range(num_target_frames)]
-                fps = 30.0
-                print(f"Downsampled high-FPS video to 30 FPS ({len(video_frames)} frames).")
-
-        # Subsample frames if excessively long (e.g. long webcam recordings) to prevent CPU freeze/timeout
-        max_process_frames = 120
-        if len(video_frames) > max_process_frames:
-            step = len(video_frames) / float(max_process_frames)
-            indices = [int(i * step) for i in range(max_process_frames)]
-            video_frames = [video_frames[i] for i in indices]
-            print(f"Subsampled long video ({len(video_frames)} frames) for fast inference.")
+        if fps is None or fps <= 0 or np.isnan(fps):
+            fps = 25.0
 
         pose_results = predict_pose(video_frames, pose_models)
 

visualize_pose.py

@@ -361,6 +361,7 @@ def render_keypoints_video(
                 "-pix_fmt", "rgb24",
                 "-r", str(fps),
                 "-i", "-",
+                "-r", str(fps),
                 "-c:v", "libx264",
                 "-preset", "fast",
                 "-crf", "18",         # Near lossless visual quality