init

app.py

@@ -60,9 +60,86 @@
 # st.info("Author ©️ : wissem karous ")
 # st.info("Made with ❤️  ")
 
+///////////////////
+# import streamlit as st
+# import os
+# from utils.demo import load_video, ctc_decode
+# from utils.two_stream_infer import load_model
+# from scripts.extract_lip_coordinates import generate_lip_coordinates
+# import options as opt
+
+# st.set_page_config(layout="wide")
+
+# model = load_model()
+
+# st.title("Lipreading final year project Demo")
+
+# st.info(
+#     "The inference speed is very slow on Huggingface spaces due to it being processed entirely on CPU ",
+#     icon="ℹ️",
+# )
+
+# # Generating a list of options or videos
+# options = sorted(os.listdir(os.path.join("app_input")))  # Ensure the list is sorted
+# selected_video = st.selectbox("Choose video", options)
+
+# # Find the index of the selected video and calculate the index of the next video
+# selected_index = options.index(selected_video)
+# next_video_index = (selected_index + 1) % len(options)  # Ensures looping back to start
+# next_video = options[next_video_index]
+
+# col1, col2 = st.columns(2)
+
+# # Function to display video in a column
+# def display_video(column, video_path, video_name):
+#     os.system(f"ffmpeg -i {video_path} -vcodec libx264 {video_name}.mp4 -y")
+#     video = open(f"{video_name}.mp4", "rb")
+#     video_bytes = video.read()
+#     column.video(video_bytes)
+
+# # Displaying the selected video in the first column
+# with col1:
+#     file_path = os.path.join("app_input", selected_video)
+#     video_name = selected_video.split(".")[0]
+#     display_video(col1, file_path, video_name)
+# # Displaying the next video in the second column
+# with col2:
+#     st.info("Expected Result !")
+#     next_file_path = os.path.join("app_input", next_video)
+#     next_video_name = next_video.split(".")[0]
+#     display_video(col2, next_file_path, next_video_name)
+
+
+# # Assuming further processing (like generating predictions) is only intended for the first (selected) video
+# with col1, st.spinner("Processing video..."):
+#     video, img_p, files = load_video(f"{video_name}.mp4", opt.device)
+#     coordinates = generate_lip_coordinates(f"{video_name}_samples")
+#     # Assuming 'frames_generated' and 'coordinates_generated' are used for control flow or further processing
+#     frames_generated = True
+#     coordinates_generated = True
+#     if frames_generated and coordinates_generated:
+#         st.markdown(f"Frames Generated for {video_name}:\n{files}")
+#         st.markdown(f"Coordinates Generated for {video_name}:\n{coordinates}")
+
+# with col2:
+#     st.info("Ready to make prediction!")
+#     generate = st.button("Generate")
+#     if generate:
+#         with st.spinner("Generating..."):
+#             y = model(
+#                 video[None, ...].to(opt.device),
+#                 coordinates[None, ...].to(opt.device),
+#             )
+#             txt = ctc_decode(y[0])
+#             st.text(txt[-1])
+
+# st.info("Author ©️ : Wissem Karous ")
+# st.info("Made with ❤️")
+/////////////////////
 
 import streamlit as st
 import os
+import cv2
 from utils.demo import load_video, ctc_decode
 from utils.two_stream_infer import load_model
 from scripts.extract_lip_coordinates import generate_lip_coordinates
@@ -72,7 +149,7 @@ st.set_page_config(layout="wide")
 
 model = load_model()
 
-st.title("Lipreading final year project Demo")
+st.title("Lipreading Final Year Project Demo")
 
 st.info(
     "The inference speed is very slow on Huggingface spaces due to it being processed entirely on CPU ",
@@ -90,18 +167,27 @@ next_video = options[next_video_index]
 
 col1, col2 = st.columns(2)
 
-# Function to display video in a column
+# Function to display video in a column with resizing
 def display_video(column, video_path, video_name):
-    os.system(f"ffmpeg -i {video_path} -vcodec libx264 {video_name}.mp4 -y")
-    video = open(f"{video_name}.mp4", "rb")
-    video_bytes = video.read()
-    column.video(video_bytes)
+    cap = cv2.VideoCapture(video_path)
+    width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
+    height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
+    new_width = 320  # Adjust this value for desired width
+    new_height = int((new_width / width) * height)
+    while cap.isOpened():
+        ret, frame = cap.read()
+        if not ret:
+            break
+        frame = cv2.resize(frame, (new_width, new_height))
+        column.image(frame, channels="BGR")
+    cap.release()
 
 # Displaying the selected video in the first column
 with col1:
     file_path = os.path.join("app_input", selected_video)
     video_name = selected_video.split(".")[0]
     display_video(col1, file_path, video_name)
+
 # Displaying the next video in the second column
 with col2:
     st.info("Expected Result !")
@@ -109,7 +195,6 @@ with col2:
     next_video_name = next_video.split(".")[0]
     display_video(col2, next_file_path, next_video_name)
 
-
 # Assuming further processing (like generating predictions) is only intended for the first (selected) video
 with col1, st.spinner("Processing video..."):
     video, img_p, files = load_video(f"{video_name}.mp4", opt.device)
@@ -135,3 +220,4 @@ with col2:
 
 st.info("Author ©️ : Wissem Karous ")
 st.info("Made with ❤️")
+