Update app.py

app.py

@@ -1,11 +1,13 @@
-import streamlit as st
-import av
+from flask import Flask, request, jsonify
+import cv2
 import joblib
 import mediapipe as mp
 import numpy as np
-from streamlit_webrtc import webrtc_streamer, VideoTransformerBase
+import tempfile
 
-# Load trained model and label encoder
+app = Flask(__name__)
+
+# Load model and label encoder
 model = joblib.load("pose_classifier.joblib")
 label_encoder = joblib.load("label_encoder.joblib")
 
@@ -13,38 +15,47 @@ label_encoder = joblib.load("label_encoder.joblib")
 mp_pose = mp.solutions.pose
 pose = mp_pose.Pose()
 
-# Streamlit UI
-st.title("Live Pose Classification on Hugging Face Spaces")
-st.write("Using Streamlit WebRTC, OpenCV, and MediaPipe.")
+def predict_pose_from_image(image_bytes):
+    # Convert image bytes to numpy array
+    nparr = np.frombuffer(image_bytes, np.uint8)
+    frame = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
+
+    if frame is None:
+        return None, "Invalid image"
+
+    # Convert to RGB
+    img_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
 
-class PoseClassification(VideoTransformerBase):
-    def transform(self, frame):
-        img = frame.to_ndarray(format="bgr24")
+    # Run MediaPipe Pose
+    results = pose.process(img_rgb)
 
-        # Convert frame to RGB
-        img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+    if results.pose_landmarks:
+        landmarks = results.pose_landmarks.landmark
+        pose_data = [j.x for j in landmarks] + [j.y for j in landmarks] + \
+                    [j.z for j in landmarks] + [j.visibility for j in landmarks]
 
-        # Process frame with MediaPipe Pose
-        results = pose.process(img_rgb)
+        pose_data = np.array(pose_data).reshape(1, -1)
+        y_pred = model.predict(pose_data)
+        predicted_label = label_encoder.inverse_transform(y_pred)[0]
+        return predicted_label, None
+    else:
+        return None, "No pose detected"
 
-        if results.pose_landmarks:
-            landmarks = results.pose_landmarks.landmark
-            pose_data = [j.x for j in landmarks] + [j.y for j in landmarks] + \
-                        [j.z for j in landmarks] + [j.visibility for j in landmarks]
+@app.route('/predict-pose', methods=['POST'])
+def predict_pose():
+    if 'frame' not in request.files:
+        return jsonify({"error": "No image frame uploaded"}), 400
 
-            pose_data = np.array(pose_data).reshape(1, -1)
+    file = request.files['frame']
+    img_bytes = file.read()
 
-            try:
-                y_pred = model.predict(pose_data)
-                predicted_label = label_encoder.inverse_transform(y_pred)[0]
+    label, error = predict_pose_from_image(img_bytes)
+    if error:
+        return jsonify({"error": error}), 400
 
-                # Draw label on frame
-                cv2.putText(img, f"Pose: {predicted_label}", (20, 50),
-                            cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 3)
-            except Exception as e:
-                st.warning(f"⚠️ Prediction Error: {e}")
+    return jsonify({"predicted_pose": label})
 
-        return av.VideoFrame.from_ndarray(img, format="bgr24")
+if __name__ == "__main__":
+    app.run(debug=True, port=5007)
 
-# Start WebRTC streamer
-webrtc_streamer(key="pose-classification", video_transformer_factory=PoseClassification)
+# curl -X POST -F "frame=@your_image.jpg" http://localhost:5007/predict-pose