Upload 3 files

README.md

@@ -1,13 +1,33 @@
 ---
-title: Car Counting
-emoji: 🚀
-colorFrom: purple
-colorTo: gray
+title: Đếm xe qua line
+emoji: 🚗
+colorFrom: blue
+colorTo: green
 sdk: gradio
-sdk_version: 6.8.0
+sdk_version: 4.44.0
 app_file: app.py
 pinned: false
 license: mit
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# 🚗 Đếm xe qua line (YOLOv8 + ByteTrack)
+
+Ứng dụng đếm phương tiện (person, car, bus, truck, motorcycle) khi qua đường line trong video.
+
+## Cách dùng
+
+1. Upload video
+2. Bấm **Xử lý video**
+3. Xem video kết quả với khung đếm theo từng loại
+
+## Lưu ý
+
+- Chạy trên **CPU Basic** (miễn phí). GPU HF tính phí ~0.40$/h.
+- Nên dùng video **ngắn** (< 30 giây) vì CPU xử lý chậm.
+
+## Công nghệ
+
+- **YOLOv8** (Ultralytics) - Object detection
+- **ByteTrack** - Multi-object tracking
+- **Supervision** - Line zone counting
+- **Gradio** - Web UI

app.py

@@ -0,0 +1,156 @@
+"""
+Đếm xe qua line - Deploy lên Hugging Face Spaces với Gradio
+"""
+import os
+import tempfile
+import cv2
+import numpy as np
+import gradio as gr
+import supervision as sv
+from ultralytics import YOLO
+
+# Load model (yolov8n nhẹ cho CPU miễn phí)
+model = YOLO("yolov8n.pt")
+CLASS_NAMES_DICT = model.model.names
+
+SELECTED_CLASS_NAMES = ['person', 'bus', 'motorcycle', 'car', 'truck']
+SELECTED_CLASS_IDS = [
+    {value: key for key, value in CLASS_NAMES_DICT.items()}[name]
+    for name in SELECTED_CLASS_NAMES
+]
+
+# Annotators (tạo 1 lần, dùng lại)
+box_annotator = sv.BoxAnnotator(thickness=4)
+label_annotator = sv.LabelAnnotator(text_thickness=2, text_scale=1.5, text_color=sv.Color.BLACK)
+trace_annotator = sv.TraceAnnotator(thickness=4, trace_length=50)
+line_zone_annotator = sv.LineZoneAnnotator(
+    thickness=4, text_thickness=2, text_scale=2,
+    display_in_count=False, display_out_count=False
+)
+
+
+def process_video(video_path):
+    """Xử lý video: đếm xe qua line và trả về video đã annotate."""
+    if video_path is None:
+        return None
+
+    # Gradio Video có thể trả về dict với key "path"
+    if isinstance(video_path, dict):
+        video_path = video_path.get("path", video_path)
+
+    # Lấy thông tin video để tính line động theo kích thước frame
+    video_info = sv.VideoInfo.from_video_path(video_path)
+    w, h = video_info.width, video_info.height
+
+    # Line ngang ở giữa khung (50% chiều cao), cách mép 5%
+    line_y = int(h * 0.5)
+    line_start = sv.Point(int(w * 0.05), line_y)
+    line_end = sv.Point(int(w * 0.95), line_y)
+
+    line_zone = sv.LineZone(start=line_start, end=line_end)
+    byte_tracker = sv.ByteTrack(
+        track_activation_threshold=0.25,
+        lost_track_buffer=30,
+        minimum_matching_threshold=0.8,
+        frame_rate=video_info.fps or 30,
+        minimum_consecutive_frames=3
+    )
+    byte_tracker.reset()
+
+    previous_positions = {}
+    class_counts = {name: 0 for name in SELECTED_CLASS_NAMES}
+    crossed_ids = set()
+
+    def callback(frame: np.ndarray, index: int) -> np.ndarray:
+        nonlocal previous_positions, class_counts, crossed_ids
+
+        results = model(frame, verbose=False)[0]
+        detections = sv.Detections.from_ultralytics(results)
+        detections = detections[np.isin(detections.class_id, SELECTED_CLASS_IDS)]
+        detections = byte_tracker.update_with_detections(detections)
+
+        # Đếm theo loại khi qua line
+        if detections.tracker_id is not None:
+            xyxy = detections.xyxy
+            for i in range(len(detections)):
+                tid = int(detections.tracker_id[i])
+                cls_id = int(detections.class_id[i])
+                cls_name = CLASS_NAMES_DICT[cls_id]
+                cx = (xyxy[i, 0] + xyxy[i, 2]) / 2
+                cy = (xyxy[i, 1] + xyxy[i, 3]) / 2
+
+                if tid in previous_positions:
+                    py = previous_positions[tid]
+                    if py < line_y and cy > line_y and (tid, 'out') not in crossed_ids:
+                        crossed_ids.add((tid, 'out'))
+                        class_counts[cls_name] = class_counts.get(cls_name, 0) + 1
+                    elif py > line_y and cy < line_y and (tid, 'in') not in crossed_ids:
+                        crossed_ids.add((tid, 'in'))
+                        class_counts[cls_name] = class_counts.get(cls_name, 0) + 1
+                previous_positions[tid] = cy
+
+        labels = [
+            f"#{tid} {CLASS_NAMES_DICT[cid]} {conf:0.2f}"
+            for conf, cid, tid in zip(
+                detections.confidence, detections.class_id, detections.tracker_id
+            )
+        ]
+
+        annotator_frame = frame.copy()
+        annotator_frame = trace_annotator.annotate(scene=annotator_frame, detections=detections)
+        annotator_frame = box_annotator.annotate(scene=annotator_frame, detections=detections)
+        annotator_frame = label_annotator.annotate(scene=annotator_frame, detections=detections, labels=labels)
+
+        line_zone.trigger(detections)
+        annotator_frame = line_zone_annotator.annotate(annotator_frame, line_counter=line_zone)
+
+        # Khung tổng đếm
+        fh, fw, _ = annotator_frame.shape
+        box_w, box_h = 280, 50 + len(SELECTED_CLASS_NAMES) * 28
+        x0, y0 = fw - box_w - 20, 20
+
+        overlay = annotator_frame.copy()
+        cv2.rectangle(overlay, (x0, y0), (x0 + box_w, y0 + box_h), (0, 0, 0), -1)
+        annotator_frame = cv2.addWeighted(overlay, 0.6, annotator_frame, 0.4, 0)
+
+        total = sum(class_counts.values())
+        cv2.putText(annotator_frame, f'Total: {total}', (x0 + 10, y0 + 30),
+                    cv2.FONT_HERSHEY_SIMPLEX, 0.8, (0, 255, 0), 2)
+        for i, cls_name in enumerate(SELECTED_CLASS_NAMES):
+            cnt = class_counts.get(cls_name, 0)
+            cv2.putText(annotator_frame, f'{cls_name.capitalize()}: {cnt}',
+                        (x0 + 10, y0 + 60 + i * 28),
+                        cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 255, 255), 2)
+
+        return annotator_frame
+
+    output_path = tempfile.NamedTemporaryFile(suffix=".mp4", delete=False).name
+    sv.process_video(
+        source_path=video_path,
+        target_path=output_path,
+        callback=callback
+    )
+    return output_path
+
+
+# Gradio UI
+with gr.Blocks(title="Đếm xe qua line", theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 🚗 Đếm xe qua line (YOLOv8 + ByteTrack)")
+    gr.Markdown("Upload video, hệ thống sẽ đếm person, car, bus, truck, motorcycle khi qua đường line giữa khung hình.")
+
+    with gr.Row():
+        video_input = gr.Video(label="Video đầu vào")
+        video_output = gr.Video(label="Video đã xử lý")
+
+    btn = gr.Button("▶️ Xử lý video")
+    btn.click(fn=process_video, inputs=video_input, outputs=video_output)
+
+    gr.Markdown("""
+    ### Lưu ý
+    - Line đếm nằm ngang ở **giữa khung hình** (50% chiều cao)
+    - Chạy trên CPU miễn phí → nên dùng video **ngắn** (< 30 giây) để tránh chờ lâu
+    - Model: YOLOv8n (nhẹ cho CPU)
+    """)
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,5 @@
+gradio>=4.0.0
+ultralytics>=8.3.0
+supervision[assets]>=0.24.0
+opencv-python-headless>=4.8.0
+numpy>=1.24.0