Update app.py

app.py

@@ -7,8 +7,9 @@ import time
 from dataclasses import dataclass
 import zipfile
 import logging
-import av
-from streamlit_webrtc import webrtc_streamer, VideoProcessorBase, WebRtcMode
+import cv2
+from PIL import Image
+import numpy as np
 
 # Logging setup
 logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s")
@@ -189,34 +190,6 @@ def zip_files(files, zip_name):
             zipf.write(file, os.path.basename(file))
     return zip_name
 
-# Video Processor for WebRTC
-class CameraProcessor(VideoProcessorBase):
-    def __init__(self):
-        self.snapshot = None
-        self.frames = []
-        self.recording = False
-
-    def recv(self, frame):
-        from PIL import Image
-        img = frame.to_image()
-        self.snapshot = img
-        if self.recording:
-            self.frames.append(frame.to_ndarray(format="bgr24"))
-        return av.VideoFrame.from_image(img)
-
-    def capture_frame(self):
-        from PIL import Image
-        return self.snapshot
-
-    def capture_video(self, duration=10):
-        self.recording = True
-        self.frames = []
-        start_time = time.time()
-        while time.time() - start_time < duration and self.recording:
-            time.sleep(0.033)  # ~30 FPS
-        self.recording = False
-        return self.frames
-
 # Main App
 st.title("SFT Tiny Titans 🚀 (Dual Cam Action!)")
 
@@ -231,7 +204,6 @@ def update_gallery():
             for idx, file in enumerate(media_files[:4]):
                 with cols[idx % 2]:
                     if file.endswith(".png"):
-                        from PIL import Image
                         st.image(Image.open(file), caption=file.split('/')[-1], use_container_width=True)
                     elif file.endswith(".mp4"):
                         st.video(file)
@@ -277,82 +249,82 @@ with tab1:
 
 with tab2:
     st.header("Camera Snap 📷 (Dual Live Feed!)")
+    caps = {0: cv2.VideoCapture(0), 1: cv2.VideoCapture(1)}
     cols = st.columns(2)
-    processors = {}
     for i in range(2):
         with cols[i]:
             st.subheader(f"Camera {i}")
-            key = f"camera_{i}"
-            processors[key] = webrtc_streamer(
-                key=key,
-                mode=WebRtcMode.SENDRECV,
-                video_processor_factory=CameraProcessor,
-                frontend_rtc_configuration={"iceServers": [{"urls": ["stun:stun.l.google.com:19302"]}]}
-            )
+            if caps[i].isOpened():
+                ret, frame = caps[i].read()
+                if ret:
+                    frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+                    st.image(frame_rgb, caption=f"Live Feed Cam {i}", use_container_width=True)
+                else:
+                    st.warning(f"Camera {i} failed to read frame!")
+                    logger.error(f"Failed to read frame from Camera {i}")
+            else:
+                st.warning(f"Camera {i} not detected!")
+                logger.error(f"Camera {i} not opened")
+
             if st.button(f"Capture Frame 📸 Cam {i}", key=f"snap_{i}"):
                 logger.info(f"Capturing frame from Camera {i}")
                 try:
-                    if processors[key].video_processor:
-                        snapshot = processors[key].video_processor.capture_frame()
-                        if snapshot:
-                            filename = generate_filename(i)
-                            snapshot.save(filename)
-                            st.image(snapshot, caption=filename, use_container_width=True)
-                            logger.info(f"Saved snapshot: {filename}")
-                            if 'captured_images' not in st.session_state:
-                                st.session_state['captured_images'] = []
-                            st.session_state['captured_images'].append(filename)
-                            update_gallery()
-                        else:
-                            st.error("No frame captured!")
-                            logger.error("No frame available for snapshot")
+                    ret, frame = caps[i].read()
+                    if ret:
+                        frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+                        img = Image.fromarray(frame_rgb)
+                        filename = generate_filename(i)
+                        img.save(filename)
+                        st.image(img, caption=filename, use_container_width=True)
+                        logger.info(f"Saved snapshot: {filename}")
+                        if 'captured_images' not in st.session_state:
+                            st.session_state['captured_images'] = []
+                        st.session_state['captured_images'].append(filename)
+                        update_gallery()
                     else:
-                        st.error("Camera processor not initialized!")
-                        logger.error("Processor not ready for Camera {i}")
+                        st.error("Failed to capture frame!")
+                        logger.error(f"No frame captured from Camera {i}")
                 except Exception as e:
                     st.error(f"Frame capture failed: {str(e)}")
                     logger.error(f"Error capturing frame: {str(e)}")
+
             if st.button(f"Capture Video 🎥 Cam {i}", key=f"rec_{i}"):
                 logger.info(f"Capturing 10s video from Camera {i}")
                 try:
-                    if processors[key].video_processor:
-                        frames = processors[key].video_processor.capture_video()
-                        if frames:
-                            mp4_filename = generate_filename(i, "mp4")
-                            with av.open(mp4_filename, "w") as container:
-                                stream = container.add_stream("h264", rate=30)
-                                stream.width = frames[0].shape[1]
-                                stream.height = frames[0].shape[0]
-                                for frame in frames:
-                                    av_frame = av.VideoFrame.from_ndarray(frame, format="bgr24")
-                                    for packet in stream.encode(av_frame):
-                                        container.mux(packet)
-                                for packet in stream.encode():
-                                    container.mux(packet)
-                            st.video(mp4_filename)
-                            logger.info(f"Saved video: {mp4_filename}")
-                            # Slice into 10 frames
-                            sliced_images = []
-                            step = max(1, len(frames) // 10)
-                            for j in range(0, len(frames), step):
-                                if len(sliced_images) < 10:
-                                    img = Image.fromarray(frames[j][:, :, ::-1])  # BGR to RGB
-                                    img_filename = generate_filename(f"{i}_{len(sliced_images)}")
-                                    img.save(img_filename)
-                                    sliced_images.append(img_filename)
-                                    st.image(img, caption=img_filename, use_container_width=True)
-                            st.session_state['captured_images'] = st.session_state.get('captured_images', []) + sliced_images
-                            logger.info(f"Sliced video into {len(sliced_images)} images")
-                            update_gallery()
-                        else:
-                            st.error("No frames recorded!")
-                            logger.error("No frames captured during video recording")
-                    else:
-                        st.error("Camera processor not initialized!")
-                        logger.error("Processor not ready for Camera {i}")
+                    fourcc = cv2.VideoWriter_fourcc(*'mp4v')
+                    mp4_filename = generate_filename(i, "mp4")
+                    out = cv2.VideoWriter(mp4_filename, fourcc, 30.0, (int(caps[i].get(3)), int(caps[i].get(4))))
+                    frames = []
+                    start_time = time.time()
+                    while time.time() - start_time < 10:
+                        ret, frame = caps[i].read()
+                        if ret:
+                            frames.append(frame)
+                            out.write(frame)
+                        time.sleep(0.033)  # ~30 FPS
+                    out.release()
+                    st.video(mp4_filename)
+                    logger.info(f"Saved video: {mp4_filename}")
+                    # Slice into 10 frames
+                    sliced_images = []
+                    step = max(1, len(frames) // 10)
+                    for j in range(0, len(frames), step):
+                        if len(sliced_images) < 10:
+                            frame_rgb = cv2.cvtColor(frames[j], cv2.COLOR_BGR2RGB)
+                            img = Image.fromarray(frame_rgb)
+                            img_filename = generate_filename(f"{i}_{len(sliced_images)}")
+                            img.save(img_filename)
+                            sliced_images.append(img_filename)
+                            st.image(img, caption=img_filename, use_container_width=True)
+                    st.session_state['captured_images'] = st.session_state.get('captured_images', []) + sliced_images
+                    logger.info(f"Sliced video into {len(sliced_images)} images")
+                    update_gallery()
                 except Exception as e:
                     st.error(f"Video capture failed: {str(e)}")
                     logger.error(f"Error capturing video: {str(e)}")
+    # Release cameras after use
+    for cap in caps.values():
+        cap.release()
 
 with tab3:
     st.header("Fine-Tune Titans 🔧 (Tune Fast!)")
@@ -379,7 +351,6 @@ with tab3:
                 if st.button("Tune CV 🔄"):
                     logger.info("Initiating CV fine-tune")
                     try:
-                        from PIL import Image
                         images = [Image.open(img) for img in captured_images]
                         st.session_state['builder'].fine_tune(images, texts)
                         st.success("CV polished! 🎉")
@@ -410,7 +381,6 @@ with tab4:
                 if st.button("Run CV Demo ▶️"):
                     logger.info("Running CV image set demo")
                     try:
-                        from PIL import Image
                         images = [Image.open(img) for img in captured_images[:10]]
                         prompts = ["Neon " + os.path.basename(img).split('.')[0] for img in captured_images[:10]]
                         generated_images = []