Update app.py

app.py

@@ -3,21 +3,41 @@ import subprocess
 import gradio as gr
 from gradio_imageslider import ImageSlider
 from PIL import Image
-import glob
-import tempfile
+import cv2
 
 # Output folder
 OUTPUT_DIR = "output"
 os.makedirs(OUTPUT_DIR, exist_ok=True)
 
+def generate_slider_from_video(video_path, max_frames=30):
+    """
+    Extract up to max_frames evenly spaced frames from the video
+    to display in the ImageSlider.
+    """
+    cap = cv2.VideoCapture(video_path)
+    frames = []
+    total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
+    step = max(1, total_frames // max_frames)
+    idx = 0
+    while True:
+        ret, frame = cap.read()
+        if not ret:
+            break
+        if idx % step == 0:
+            frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+            frames.append(Image.fromarray(frame_rgb))
+        idx += 1
+    cap.release()
+    return frames
+
 def process_video(video_file):
     """
-    Run DepthAnythingV2 run_video.py CLI to process the uploaded MP4 video
-    and return both a slider view (per-frame depth map) and the output video.
+    Call run_video.py to process the uploaded video and return
+    both a slider view and the output depth video.
     """
     video_path = video_file.name
 
-    # Call run_video.py using subprocess
+    # Run DepthAnythingV2 CLI
     cmd = [
         "python", "run_video.py",
         "--encoder", "vitb",
@@ -28,26 +48,22 @@ def process_video(video_file):
     ]
     subprocess.run(cmd, check=True)
 
-    # run_video.py output video filename: <input>_depth.mp4
+    # Construct output video path (run_video.py convention)
     base_name = os.path.basename(video_path)
     output_video = os.path.join(OUTPUT_DIR, base_name.replace(".mp4", "_depth.mp4"))
 
-    # Prepare slider images for per-frame preview
-    # Assume run_video.py saves frames as PNG in output folder
-    slider_images = []
-    frame_files = sorted(glob.glob(os.path.join(OUTPUT_DIR, "*.png")))
-    for f in frame_files:
-        slider_images.append(Image.open(f))
+    # Generate slider images from output video
+    slider_images = generate_slider_from_video(output_video, max_frames=30)
 
     return slider_images, output_video
 
 # Gradio UI
 with gr.Blocks() as demo:
     gr.Markdown("# Depth Anything V2 - Video Demo")
-    gr.Markdown("Upload an MP4 video, and the system will automatically generate a DepthMap video.")
+    gr.Markdown("Upload an MP4 video and get a DepthMap video automatically.")
 
     video_input = gr.File(label="Upload MP4", file_types=['.mp4'])
-    depth_slider = ImageSlider(label="Depth Map Slider")  # Slider view of frames
+    depth_slider = ImageSlider(label="Depth Map Slider")  # per-frame preview
     video_output = gr.Video(label="DepthMap Video")
     submit = gr.Button("Render DepthMap")