Create app.py

app.py

@@ -0,0 +1,125 @@
+import gradio as gr
+import cv2
+import os
+import uuid
+import mediapipe as mp
+import pandas as pd
+import threading
+import time
+
+# Setup
+OUTPUT_DIR = "captured_frames"
+os.makedirs(OUTPUT_DIR, exist_ok=True)
+df = pd.DataFrame(columns=["filename", "caption", "pose_coords"])
+
+pose = mp.solutions.pose.Pose()
+state = {
+    "video_path": None,
+    "cap": None,
+    "frame": None,
+    "play": False,
+    "thread": None
+}
+
+def load_video(video_file):
+    if state["cap"]:
+        state["cap"].release()
+    state["video_path"] = video_file.name
+    state["cap"] = cv2.VideoCapture(state["video_path"])
+    state["play"] = False
+    return "Video loaded. Use play/pause to navigate."
+
+def play_video():
+    def play():
+        while state["cap"].isOpened() and state["play"]:
+            ret, frame = state["cap"].read()
+            if not ret:
+                break
+            state["frame"] = frame
+            time.sleep(0.1)  # ~10 FPS
+
+    state["play"] = True
+    state["thread"] = threading.Thread(target=play)
+    state["thread"].start()
+    return "Playing video..."
+
+def pause_video():
+    state["play"] = False
+    return "Video paused."
+
+def show_frame():
+    if state["frame"] is not None:
+        return state["frame"][:, :, ::-1]  # BGR to RGB
+    return None
+
+def capture_frame(caption):
+    if state["frame"] is None:
+        return "No frame to capture.", None
+
+    frame = state["frame"]
+    filename = f"{uuid.uuid4().hex[:8]}.jpg"
+    path = os.path.join(OUTPUT_DIR, filename)
+    cv2.imwrite(path, frame)
+
+    frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+    results = pose.process(frame_rgb)
+
+    coords = []
+    if results.pose_landmarks:
+        for lm in results.pose_landmarks.landmark:
+            coords.append((round(lm.x, 5), round(lm.y, 5), round(lm.z, 5)))
+
+    global df
+    df = pd.concat([df, pd.DataFrame([{
+        "filename": filename,
+        "caption": caption,
+        "pose_coords": coords
+    }])], ignore_index=True)
+
+    return f"Frame captured: {filename}", frame[:, :, ::-1]
+
+def download_csv():
+    csv_path = os.path.join(OUTPUT_DIR, "pose_dataset.csv")
+    df.to_csv(csv_path, index=False)
+    return csv_path
+
+def reset_all():
+    df.drop(df.index, inplace=True)
+    for f in os.listdir(OUTPUT_DIR):
+        os.remove(os.path.join(OUTPUT_DIR, f))
+    if state["cap"]:
+        state["cap"].release()
+    state.update({"video_path": None, "cap": None, "frame": None, "play": False})
+    return "App reset.", None
+
+with gr.Blocks() as app:
+    gr.Markdown("## 🏹 Archery Dataset Tool (Play, Pause, Pose + Annotation)")
+
+    video_input = gr.Video(label="Upload Video")
+    load_btn = gr.Button("Load Video")
+
+    with gr.Row():
+        play_btn = gr.Button("▶️ Play")
+        pause_btn = gr.Button("⏸️ Pause")
+        show_btn = gr.Button("🔁 Show Current Frame")
+
+    image_box = gr.Image(label="Current Frame")
+    caption_input = gr.Textbox(label="Caption")
+    capture_btn = gr.Button("📸 Capture & Annotate Frame")
+
+    download_btn = gr.Button("📥 Download CSV")
+    csv_file = gr.File(label="Pose Dataset CSV")
+
+    reset_btn = gr.Button("🔄 Reset")
+    status = gr.Textbox(label="Status")
+
+    # Event handlers
+    load_btn.click(load_video, inputs=video_input, outputs=status)
+    play_btn.click(play_video, outputs=status)
+    pause_btn.click(pause_video, outputs=status)
+    show_btn.click(show_frame, outputs=image_box)
+    capture_btn.click(capture_frame, inputs=caption_input, outputs=[status, image_box])
+    download_btn.click(download_csv, outputs=csv_file)
+    reset_btn.click(reset_all, outputs=[status, image_box])
+
+app.launch()