← Models

CV / YOLOv8

YOLOv8n_pose

YOLOv8n_Pose Human Keypoint Detection Series lightweight version, compact and high-efficiency, tailored for real-time pose estimation and keypoint detection tasks, perfectly suited for low-power edge deployment.

23 downloads
Size
~7.4MB
Memory
1GB+
Precision
INT8

Choose the device you're using, the set up guide and documentation will update accordingly.

Getting Started

Docker
sudo docker run -it --rm --pull always --runtime=nvidia \
  --network host ghcr.io/seeed-studio/vllm:latest-rk3588 \
  vllm serve yolov8n_pose

REST API

Use the REST API to run inference. Copy the commands below.

Curl
curl -X POST "http://127.0.0.1:8000/api/models/yolov8n_obb/predict" -F "realtime=true"
# Or without file parameters
curl -X POST "http://127.0.0.1:8000/api/models/yolov8_pose/predict"
Python
import requests
import json
resp = requests.post(
    "http://127.0.0.1:8000/api/models/yolov8_pose/predict",
    json={},
    timeout=30
)
result = resp.json()
print(json.dumps(result, indent=2, ensure_ascii=False))

Model Details

Quick Start

1. Install Docker

Run the following commands on the development board to install Docker:

bash
# Download installation script
curl -fsSL https://get.docker.com -o get-docker.sh
# Install using Aliyun mirror source
sudo sh get-docker.sh --mirror Aliyun
# Start Docker and enable auto-start on boot
sudo systemctl enable docker
sudo systemctl start docker

2. Run the Project (One command, Web preview)

This project supports preview via Web Browser. The program automatically serves a web interface for real-time detection or local video analysis.

Step A: Configure Display Permissions (Optional)

If you have a monitor connected and want to see the window locally:

bash
xhost +local:docker

Step B: One-click Run

bash
sudo docker run --rm --privileged --net=host \
    -e PYTHONUNBUFFERED=1 \
    -e RKNN_LOG_LEVEL=0 \
    --device /dev/video1:/dev/video1 \
    --device /dev/dri/renderD129:/dev/dri/renderD129 \
    -v /proc/device-tree/compatible:/proc/device-tree/compatible \
    ghcr.io/seeed-projects/recomputer-rk-cv/rk3576-yolov8-pose:latest \
    python3 web_detection.py --model_path model/yolov8n_pose.rknn --video video/test.mp4

Access via: http://<Board_IP>:8000

Note: If you want to test with a local video instead of a camera, use the --video parameter:

bash
sudo docker run --rm --privileged --net=host \
    -e PYTHONUNBUFFERED=1 \
    -e RKNN_LOG_LEVEL=0 \
    -v $(pwd)/video:/app/video \
    --device /dev/dri/renderD129:/dev/dri/renderD129 \
    -v /proc/device-tree/compatible:/proc/device-tree/compatible \
    ghcr.io/seeed-projects/recomputer-rk-cv/rk3576-yolov8-pose:latest \
    python3 web_detection.py --model_path model/yolov8n_pose.rknn --video video/test.mp4

🔌 API Documentation

This project provides RESTful interfaces compatible with the Ultralytics Cloud API standard, supporting pose estimation via image, video uploads or direct camera calls.

1. Model Inference Interface (Predict)

Endpoint: POST /api/models/yolov8_pose/predict

Request Parameters (Multipart/Form-Data):

  • file: (Optional) Image file to be detected.
  • video: (Optional) MP4 video file to be detected.
  • timestamp: (Optional) Timestamp in the video file (seconds), returns detection results for the frame at that point. Default is 0.
  • realtime: (Optional) Boolean. If true or if no file/video parameters are provided, returns detection results for the current camera frame.
  • conf: (Optional) Confidence threshold for a single request, range 0.0-1.0.
  • iou: (Optional) NMS IOU threshold for a single request, range 0.0-1.0.

Usage Examples:

1. Image Detection:

bash
curl -X POST "http://127.0.0.1:8000/api/models/yolov8_pose/predict" -F "file=@/home/cat/001.jpg"

2. Video Specific Frame Detection:

bash
curl -X POST "http://127.0.0.1:8000/api/models/yolov8_pose/predict" -F "video=@/home/cat/test.mp4" -F "timestamp=5.5"

3. Get Current Camera Frame Detection:

bash
curl -X POST "http://127.0.0.1:8000/api/models/yolov8_pose/predict" -F "realtime=true"
# Or without file parameters
curl -X POST "http://127.0.0.1:8000/api/models/yolov8_pose/predict"

Response Format (JSON):

json
{
  "success": true,
  "source": "video frame at 5.5s",
  "predictions": [
    {
      "class": "person",
      "confidence": 0.92,
      "box": { "x1": 100, "y1": 200, "x2": 300, "y2": 500 },
      "keypoints": [
        { "x": 150.5, "y": 210.2, "conf": 0.88 },
        { "x": 160.1, "y": 205.5, "conf": 0.91 },
        { "x": 140.2, "y": 206.1, "conf": 0.85 },
        { "x": 170.8, "y": 215.3, "conf": 0.89 },
        { "x": 130.4, "y": 218.7, "conf": 0.82 },
        { "x": 180.2, "y": 250.4, "conf": 0.95 },
        { "x": 120.5, "y": 255.1, "conf": 0.93 },
        { "x": 190.6, "y": 300.2, "conf": 0.87 },
        { "x": 110.3, "y": 305.8, "conf": 0.86 },
        { "x": 200.1, "y": 350.5, "conf": 0.81 },
        { "x": 100.9, "y": 355.2, "conf": 0.80 },
        { "x": 175.4, "y": 400.1, "conf": 0.90 },
        { "x": 125.6, "y": 405.3, "conf": 0.88 },
        { "x": 185.2, "y": 450.6, "conf": 0.85 },
        { "x": 115.8, "y": 455.4, "conf": 0.84 },
        { "x": 195.7, "y": 500.2, "conf": 0.79 },
        { "x": 105.3, "y": 505.7, "conf": 0.78 }
      ]
    }
  ],
  "image": { "width": 1280, "height": 720 }
}

2. Local Video Analysis

  • Upload Video: POST /api/video/upload
  • List Videos: GET /api/video/list
  • Analyze Video: POST /api/video/analyze
  • Check Status: GET /api/video/status
  • Download Result: GET /api/video/download/{filename}

3. System Configuration Interface (Config)

Used to dynamically adjust thresholds for real-time video streams and default inference.

Get Current Configuration

  • Endpoint: GET /api/config
  • Response: {"obj_thresh": 0.25, "nms_thresh": 0.45, "camera_id": 1, "video_path": null}

Update System Configuration

  • Endpoint: POST /api/config
  • Request Body (JSON): {"obj_thresh": 0.3, "nms_thresh": 0.5}
  • Response: {"status": "success"}

4. Real-time Video Stream Interface (Video Feed)

Get real-time MJPEG video stream with pose skeletons drawn, can be directly embedded in HTML <img> tags.

  • Endpoint: GET /api/video_feed
  • Example Usage: <img src="http://<Board_IP>:8000/api/video_feed">

🛠️ Developer Guide (Production Recommendations)

Code Description

  • web_detection.py:
    • Web API: Integrates FastAPI, supporting MJPEG streaming output, file upload, and video analysis.
    • RKNN Inference: Encapsulates RKNN initialization, model loading, and multi-core inference logic.
    • Mode Switching: Dynamically handles UI layout based on input parameters (--camera_id, --video_path).
  • py_utils/pose_utils.py:
    • Post-processing: YOLOv8-Pose specific Box decoding, NMS logic, and Keypoint extraction.
    • Visualization: Draws bounding boxes and pose skeletons with customized colors.

Modifying Models

  1. Place the trained and converted .rknn model into the model/ directory.
  2. Add the --model_path argument to the running command to point to the new model.