app.py

import os
import cv2
import mediapipe as mp
import numpy as np
import tempfile
import warnings
import gradio as gr

# --- Silence unnecessary logs and warnings ---
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
os.environ['GLOG_minloglevel'] = '2'
warnings.filterwarnings("ignore", category=UserWarning)
warnings.filterwarnings("ignore", message="SymbolDatabase.GetPrototype")

# --- Main processing function ---
def process_pose(video_path, mode, min_det_conf, min_track_conf, progress=gr.Progress()):
    try:
        progress(0, desc="Initializing pose model...")
        mp_drawing = mp.solutions.drawing_utils
        mp_pose = mp.solutions.pose

        # Temporary output file
        temp_dir = tempfile.mkdtemp()
        output_path = os.path.join(temp_dir, "output.mp4")

        cap = cv2.VideoCapture(video_path)
        total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
        width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
        height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
        fps = cap.get(cv2.CAP_PROP_FPS)

        if fps == 0 or total_frames == 0:
            raise ValueError("Could not read the video or it is empty.")

        fourcc = cv2.VideoWriter_fourcc(*'mp4v')
        out = cv2.VideoWriter(output_path, fourcc, fps, (width, height))

        with mp_pose.Pose(
            min_detection_confidence=min_det_conf,
            min_tracking_confidence=min_track_conf
        ) as pose:

            frame_idx = 0
            while cap.isOpened():
                ret, frame = cap.read()
                if not ret:
                    break

                frame_idx += 1
                progress(frame_idx / total_frames, desc=f"Processing frame {frame_idx}/{total_frames}")

                image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
                results = pose.process(image)
                image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)

                if mode == "Pose on original video":
                    output_frame = image.copy()
                    mp_drawing.draw_landmarks(
                        output_frame,
                        results.pose_landmarks,
                        mp_pose.POSE_CONNECTIONS,
                        mp_drawing.DrawingSpec(color=(245,117,66), thickness=2, circle_radius=2),
                        mp_drawing.DrawingSpec(color=(245,66,230), thickness=2, circle_radius=2)
                    )

                else:  # Pose only (black background)
                    mask = np.zeros_like(image)
                    mp_drawing.draw_landmarks(
                        mask,
                        results.pose_landmarks,
                        mp_pose.POSE_CONNECTIONS,
                        mp_drawing.DrawingSpec(color=(245,117,66), thickness=2, circle_radius=2),
                        mp_drawing.DrawingSpec(color=(245,66,230), thickness=2, circle_radius=2)
                    )
                    output_frame = cv2.addWeighted(np.zeros_like(image), 1, mask, 0.8, 0)

                out.write(output_frame)

        cap.release()
        out.release()
        progress(1, desc="Completed ✅")
        return output_path

    except Exception as e:
        return f"❌ Error during processing: {e}"

# --- Custom HTML for warnings and explanations ---
warning_html = """
<div style="text-align:center; color:red; font-weight:bold;">
⚠️ Reminder: Video must be under 5 MB due to CPU processing time.<br>
⚠️ Processing long videos may take a considerable amount of time.
</div>
"""

param_info_html = """
<div style="border: 2px solid #4CAF50; padding: 10px; border-radius: 8px; margin:10px; background-color:#f9f9f9;">
<b>Parameters:</b><br>
- <b>min_detection_confidence:</b> Minimum confidence for the model to detect a pose.<br>
- <b>min_tracking_confidence:</b> Minimum confidence for the model to track the pose across frames.
</div>
"""

# --- Gradio Interface ---
iface = gr.Interface(
    fn=process_pose,
    inputs=[
        gr.Video(label="🎥 Upload your video", sources=["upload"], elem_id="video_upload"),
        gr.Radio(
            ["Pose on original video", "Pose only (black background)"],
            label="Output mode",
            value="Pose on original video"
        ),
        gr.Slider(0.0, 1.0, value=0.5, label="min_detection_confidence"),
        gr.Slider(0.0, 1.0, value=0.5, label="min_tracking_confidence"),
    ],
    outputs=gr.Video(label="📦 Processed Video"),
    title="<center>Pose Estimation - MediaPipe (CPU Optimized)</center>",
    description=warning_html + param_info_html,
)

if __name__ == "__main__":
    iface.launch()