Update pages/Video Upload.py

pages/Video Upload.py

@@ -1,129 +1,187 @@
-import logging
-import queue
-from collections import deque
-from concurrent.futures import ThreadPoolExecutor
-
 import streamlit as st
-from streamlit_webrtc import WebRtcMode, webrtc_streamer
-
-import cv2
-from model import Predictor  # Import Predictor from your model file
-DEFAULT_WIDTH = 50
-
-import openai
-
-# Initialize the OpenAI client
-openai.api_key  ='sk-proj-GDxupB1DFvTTWBg38VyST3BlbkFJ7MdcACLwu3u0U1QvWeMb'
-
-def correct_text_gpt3(input_text):
-    prompt = f"Исправь грамматические ошибки в тексте: '{input_text}'"
-    response = openai.ChatCompletion.create(
-        model="gpt-3.5-turbo",
-        messages=[
-            {"role": "system", "content": "You are a helpful assistant that corrects grammatical errors."},
-            {"role": "user", "content": prompt}
-        ],
-        max_tokens=50,
-        n=1,
-        stop=None,
-        temperature=0.5,
-    )
-
-    corrected_text = response.choices[0].message['content'].strip()
-    return corrected_text
-
-#st.set_page_config(layout="wide")
-
-
-
-width = 50
-side = max((100 - width) / 1.2, 0.01)
-
-_, container, _ = st.columns([side, width, side])
-
-logger = logging.getLogger(__name__)
-
-class SLInference:
-    def __init__(self, config_path):
-        self.config = self.load_config(config_path)
-        self.predictor = Predictor(self.config)
-        self.input_queue = deque(maxlen=32)  # Queue to store 32 frames
-        self.pred = ''
-
-    def load_config(self, config_path):
-        import json
-        with open(config_path, 'r') as f:
-            return json.load(f)
-
-    def start(self):
-        pass  # This method can be left empty or add initialization logic
-
-    def predict(self, frames):
-        frames_resized = [cv2.resize(frame, (224, 224)) for frame in frames]
-        while len(frames_resized) < 32:
-            frames_resized.append(frames_resized[-1])
-        result = self.predictor.predict(frames_resized)
-        if result:
-            return result["labels"][0]
-        return 'no'
-
-def process_batch(inference_thread, frames, gestures):
-    gesture = inference_thread.predict(frames)
-    if gesture not in ['no', ''] and gesture not in gestures:
-        gestures.append(gesture)
-
-def main(config_path):
-    #st.set_page_config(layout="wide")
-    st.title("Sign Language Recognition Demo")
-
+from PIL import Image
+import numpy as np
+
+# Page config
+st.set_page_config(
+    page_title="GestureGuru",
+    page_icon="🤖",
+    layout="wide",
+    initial_sidebar_state="expanded"
+)
+
+# --- Custom Style ---
+st.markdown("""
+<style>
+body {
+    background-image: url('https://images.unsplash.com/photo-1498050108023-c5249f4df085?auto=format&fit=crop&w=1920&q=80');
+    background-size: cover;
+    background-attachment: fixed;
+    background-position: center;
+    font-family: 'Segoe UI', sans-serif;
+    color: #222;
+    animation: fade-in 2s ease-out;
+}
+.sidebar .sidebar-content {
+    background-color: rgba(255, 255, 255, 0.8);
+    padding: 1rem;
+    border-radius: 16px;
+    box-shadow: 0 4px 12px rgba(0,0,0,0.1);
+    backdrop-filter: blur(8px);
+}
+.hero {
+    text-align: center;
+    padding: 5rem 2rem;
+    background: linear-gradient(135deg, #8e2de2, #4a00e0);
+    border-radius: 24px;
+    color: white;
+    margin-bottom: 3rem;
+    box-shadow: 0 20px 40px rgba(0,0,0,0.1);
+    animation: fade-in 2s ease-out;
+    position: relative;
+}
+.hero h1 {
+    font-size: 3.75em;
+    font-weight: 700;
+    margin-bottom: 0.3em;
+    animation: glow-text 2s infinite alternate;
+}
+.hero p {
+    font-size: 1.3em;
+    font-weight: 300;
+    margin-top: 0.5em;
+    animation: slide-in-bottom 1.2s ease-out;
+}
+.hero img.logo {
+    width: 120px;
+    position: absolute;
+    top: 20px;
+    left: 20px;
+    border-radius: 50%;
+    border: 2px solid white;
+    box-shadow: 0 4px 12px rgba(0,0,0,0.4);
+}
+.section {
+    background: rgba(255, 255, 255, 0.95);
+    border-radius: 20px;
+    padding: 2.5rem;
+    margin-bottom: 2.5rem;
+    box-shadow: 0 10px 30px rgba(0, 0, 0, 0.06);
+    backdrop-filter: blur(4px);
+    animation: fade-in-up 1s ease-out;
+}
+.upload-section {
+    background: linear-gradient(to right, #00c6ff, #0072ff);
+    color: white;
+    padding: 2rem;
+    border-radius: 20px;
+    margin-bottom: 2rem;
+    box-shadow: 0 8px 20px rgba(0, 0, 0, 0.2);
+    animation: fade-in-up 1s ease-out;
+}
+.upload-section h3 {
+    color: white;
+}
+h3 {
+    font-size: 1.8rem;
+    color: #2e2e2e;
+    margin-bottom: 1rem;
+}
+ul, ol {
+    font-size: 1.1em;
+    line-height: 1.6;
+    color: #444;
+}
+.contact-block {
+    text-align: center;
+    padding: 3rem;
+    border-radius: 20px;
+    background: linear-gradient(to right, #ff6a00, #ee0979);
+    color: white;
+    margin-top: 3rem;
+    box-shadow: 0 10px 30px rgba(0, 0, 0, 0.1);
+    animation: fade-in 2s ease-out;
+}
+.contact-block a {
+    color: #fff;
+    text-decoration: underline;
+}
+@keyframes fade-in {
+  from { opacity: 0; }
+  to { opacity: 1; }
+}
+@keyframes fade-in-up {
+  from { opacity: 0; transform: translateY(30px); }
+  to { opacity: 1; transform: translateY(0); }
+}
+@keyframes slide-in-top {
+  0% { transform: translateY(-100px); opacity: 0; }
+  100% { transform: translateY(0); opacity: 1; }
+}
+@keyframes slide-in-bottom {
+  0% { transform: translateY(100px); opacity: 0; }
+  100% { transform: translateY(0); opacity: 1; }
+}
+@keyframes glow-text {
+  from { text-shadow: 0 0 5px #fff, 0 0 10px #bca7ff, 0 0 20px #7e57ff; }
+  to { text-shadow: 0 0 10px #fff, 0 0 20px #b388ff, 0 0 30px #7e57ff; }
+}
+</style>
+""", unsafe_allow_html=True)
+
+# --- Hero Section ---
+st.markdown("""
+<div class="hero">
+    <img src="https://cdn-icons-png.flaticon.com/512/3062/3062634.png" class="logo" alt="logo">
+    <h1>GestureGuru</h1>
+    <p>AI for Sign Language Understanding & Accessibility</p>
+</div>
+""", unsafe_allow_html=True)
+
+# --- Features ---
+st.markdown("""
+<div class="section">
+<h3>✨ Key Features</h3>
+<ul>
+  <li>⚡ Real-time AI gesture recognition</li>
+  <li>🧠 Optimized for Russian Sign Language</li>
+  <li>🖥️ Lightweight and browser-based interface</li>
+  <li>🌍 Promoting accessibility and inclusive communication</li>
+</ul>
+</div>
+""", unsafe_allow_html=True)
+
+# --- How it works ---
+st.markdown("""
+<div class="section">
+<h3>🛠 How it works</h3>
+<ol>
+  <li>Camera captures hand gestures</li>
+  <li>Frame is processed via ONNX deep learning model</li>
+  <li>Text prediction appears instantly on screen</li>
+</ol>
+</div>
+""", unsafe_allow_html=True)
+
+# --- Upload Section ---
+st.markdown("""
+<div class="upload-section">
+    <h3>🎥 Sign Language Recognition Demo</h3>
+    <p>Upload a short video clip to detect sign gestures:</p>
+</div>
+""", unsafe_allow_html=True)
+
+uploaded_file = st.file_uploader("Upload Video", type=["mp4", "avi", "mov", "gif"])
+if uploaded_file:
+    st.video(uploaded_file)
+else:
     st.warning("Please upload a video file for prediction.")
 
-    uploaded_file = st.file_uploader("Upload Video", type=["mp4", "avi", "mov", "gif"])
-
-    if uploaded_file is not None:
-        video_bytes = uploaded_file.read()
-        container.video(data=video_bytes)
-        #st.video(video_bytes)
-
-        inference_thread = SLInference(config_path)
-        inference_thread.start()
-
-        text_output = st.empty()
-
-        if st.button("Predict"):
-            import tempfile
-            tfile = tempfile.NamedTemporaryFile(delete=False)
-            tfile.write(video_bytes)
-            cap = cv2.VideoCapture(tfile.name)
-
-            gestures = []
-            frames = []
-            batch_size = 32
-
-            def process_frames(batch):
-                process_batch(inference_thread, batch, gestures)
-
-            with ThreadPoolExecutor() as executor:
-                while cap.isOpened():
-                    ret, frame = cap.read()
-                    if not ret:
-                        break
-                    frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
-                    frames.append(frame)
-                    if len(frames) == batch_size:
-                        executor.submit(process_frames, frames)
-                        frames = []
-
-                if frames:
-                    executor.submit(process_frames, frames)
-
-            cap.release()
-            text_output.markdown(f'<p style="font-size:20px"> Gestures in video: {" ".join(gestures)}</p>',
-                                 unsafe_allow_html=True)
-            st.text(correct_text_gpt3(" ".join(gestures)))
-            
-            print(gestures)
-
-if __name__ == "__main__":
-    logging.basicConfig(level=logging.INFO)
-    main("configs/config.json")
+# --- Contact block ---
+st.markdown("""
+<div class="contact-block">
+    <h3>📬 Let's collaborate!</h3>
+    <p>For demos, partnerships, or research opportunities:</p>
+    <p>Email: <a href="mailto:your@email.com">your@email.com</a></p>
+</div>
+""", unsafe_allow_html=True)