Update app.py

app.py

@@ -13,6 +13,14 @@ import spaces
 
 from depth_anything.dpt import DepthAnything
 from depth_anything.util.transform import Resize, NormalizeImage, PrepareForNet
+from moviepy.editor import *
+
+def create_video(frames, fps, type):
+    print("building video result")
+    clip = ImageSequenceClip(frames, fps=fps)
+    clip.write_videofile(type + "_result.mp4", fps=fps)
+    
+    return type + "_result.mp4"
 
 @torch.no_grad()
 def predict_depth(model, image):
@@ -82,7 +90,8 @@ def make_video(video_path, outdir='./vis_video_depth', encoder='vits'):
         #out = cv2.VideoWriter(output_path, cv2.VideoWriter_fourcc(*"avc1"), frame_rate, (output_width, frame_height))
         fourcc = cv2.VideoWriter_fourcc(*'mp4v')
         out = cv2.VideoWriter(output_path, fourcc, frame_rate, (output_width, frame_height))
-        # count=0
+        count=0
+        depth_frames = []
         while raw_video.isOpened():
             ret, raw_frame = raw_video.read()
             if not ret:
@@ -109,12 +118,16 @@ def make_video(video_path, outdir='./vis_video_depth', encoder='vits'):
             # out.write(combined_frame)
             # frame_path = os.path.join(temp_frame_dir, f"frame_{count:05d}.png")
             # cv2.imwrite(frame_path, combined_frame)
-            out.write(combined_frame)
-            # count += 1
-        
+            cv2.imwrite(f"frame{count}.jpg", combined_frame)
+            depth_frames.append(f"frame{count}.jpg")
+            count += 1
+
+        final_vid = create_video(depth_frames, frame_rate, "depth")
         raw_video.release()
         out.release()
-        return output_path
+        cv2.destroyAllWindows()
+        
+        return final_vid #output_path
 
 def loadurl(url):
     return url