upload v2

app.py

@@ -3,7 +3,8 @@ import av
 import torch
 from transformers import AutoImageProcessor, AutoModelForVideoClassification
 import streamlit as st
-
+import torch.nn as nn
+from streamlit_navigation_bar import st_navbar
 
 def read_video_pyav(container, indices):
     '''
@@ -43,61 +44,92 @@ def sample_frame_indices(clip_len, frame_sample_rate, seg_len):
     indices = np.clip(indices, start_idx, end_idx - 1).astype(np.int64)
     return indices
 
-# def sample_frame_indices2(clip_len, frame_sample_rate, seg_len):
-#     '''
-#     Description
-#     Args:
-#     Returns:
-#         indices (`List[int]`): List of sampled frame indices
-#     '''
-#     return 
-
-def classify(file):
+def victoire():
+    gif_url = "https://i.postimg.cc/rDp7xRJY/Happy-Birthday-Confetti.gif"
+    html_gif = f"""
+        <div style="display: flex; justify-content: center; align-items: center;">
+            <img src="{gif_url}" height="auto" style="margin: 0px;">
+            <img src="{gif_url}" height="auto" style="margin: 0px;">
+            <img src="{gif_url}" height="auto" style="margin: 0px;">
+            <img src="{gif_url}" height="auto" style="margin: 0px;">
+        </div>
+    """
+    st.markdown(html_gif, unsafe_allow_html=True)
+
+def classify(model_maneuver,model_Surf_notSurf,file):
     container = av.open(file)
 
     # sample 16 frames
     indices = sample_frame_indices(clip_len=16, frame_sample_rate=4, seg_len=container.streams.video[0].frames)
     video = read_video_pyav(container, indices)
 
-    if container.streams.video[0].frames < 16:
-        return 'Video trop courte'
-
     inputs = image_processor(list(video), return_tensors="pt")
 
     with torch.no_grad():
-        outputs = model(**inputs)
+        outputs = model_Surf_notSurf(**inputs)
         logits = outputs.logits
 
-    # model predicts one of the 400 Kinetics-400 classes
     predicted_label = logits.argmax(-1).item()
-    print(model.config.id2label[predicted_label])
-    
-    return model.config.id2label[predicted_label]
+    print(model_Surf_notSurf.config.id2label[predicted_label])
+
+    if model_Surf_notSurf.config.id2label[predicted_label]!='Surfing':
+        return model_Surf_notSurf.config.id2label[predicted_label]
+    else:
+        with torch.no_grad():
+            outputs = model_maneuver(**inputs)
+            logits = outputs.logits
+
+        predicted_label = logits.argmax(-1).item()
+        print(model_maneuver.config.id2label[predicted_label])
+        # st.write(f'Les labels: {model_maneuver.config.id2label}')
+        # st.write(f'répartiton des probilités {logits}')
+        # st.write(f'répartiton des probilités {nn.Softmax(dim=-1)(logits)}')
+        
+        return model_maneuver.config.id2label[predicted_label]
+
+
+model_maneuver = '2nzi/videomae-surf-analytics'
+model_Surf_notSurf = '2nzi/videomae-surf-analytics-surfNOTsurf'
+image_processor = AutoImageProcessor.from_pretrained(model_maneuver)
+model_maneuver = AutoModelForVideoClassification.from_pretrained(model_maneuver)
+model_Surf_notSurf = AutoModelForVideoClassification.from_pretrained(model_Surf_notSurf)
+
+
 
 
-model_ckpt = '2nzi/videomae-surf-analytics'
-# pipe = pipeline("video-classification", model="2nzi/videomae-surf-analytics")
-image_processor = AutoImageProcessor.from_pretrained(model_ckpt)
-model = AutoModelForVideoClassification.from_pretrained(model_ckpt)
+# Define the navigation bar and its pages
+page = st_navbar(["Home", "Documentation", "Examples", "About Us"])
 
 
 
+# Main application code
+if page == "Home":
+    st.subheader("Surf Analytics")
+    st.markdown("""
+        Bienvenue sur le projet Surf Analytics réalisé par Walid, Guillaume, Valentine, et Antoine.
+                
+        <a href="https://github.com/2nzi/M09-FinalProject-Surf-Analytics" style="text-decoration: none;">@Surf-Analytics-Github</a>.
+    """, unsafe_allow_html=True)
+    st.title("Surf Maneuver Classification")
 
+    uploaded_file = st.file_uploader("Upload a video file", type=["mp4"])
 
-st.subheader("Surf Analytics")
+    if uploaded_file is not None:
+        video_bytes = uploaded_file.read()
+        st.video(video_bytes)
+        predicted_label = classify(model_maneuver, model_Surf_notSurf, uploaded_file)
+        st.success(f"Predicted Label: {predicted_label}")
+        victoire()
 
-st.markdown("""
-    Bienvenue sur le projet Surf Analytics réalisé par Walid, Guillaume, Valentine, et Antoine.
-            
-    <a href="https://github.com/2nzi/M09-FinalProject-Surf-Analytics" style="text-decoration: none;">@Surf-Analytics-Github</a>.
-""", unsafe_allow_html=True)
 
-st.title("Surf Maneuver Classification")
+elif page == "Documentation":
+    st.title("Documentation")
+    st.markdown("Here you can add your documentation content.")
 
-uploaded_file = st.file_uploader("Upload a video file", type=["mp4", "avi", "mov"])
+elif page == "Examples":
+    st.title("Examples")
+    st.markdown("Here you can add examples related to your project.")
 
-if uploaded_file is not None:
-    video_bytes = uploaded_file.read()
-    st.video(video_bytes)
-    predicted_label = classify(uploaded_file)
-    st.success(f"Predicted Label: {predicted_label}")
+elif page == "About Us":
+    st.title("About")
+    st.markdown("Here you can add information about the project and the team.")