CV / YOLOv6

YOLOv6

reComputer RK3588

YOLOv6 COCO object detection on reComputer RK3576/RK3588 through RKNN and on reComputer R Series with Hailo-8 or Hailo-10H through HailoRT.

97 downloads

Precision

RKNN / Hailo HEF

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

Platform

Size

Inference Engine

Getting Started

Armbian 26.8.7 Gnome or newer with OTA overlayroot: Configure Docker storage before running containers to avoid overlay mount errors.

Deploy

sudo docker run --rm --privileged --net=host --device /dev/dri/renderD128:/dev/dri/renderD128 \
  -v /proc/device-tree/compatible:/proc/device-tree/compatible:ro \
  ghcr.io/seeed-projects/recomputer-rk-cv/rk3588-yolov6:latest \
  python3 web_detection.py --platform rk3588 --model_path model/yolov6n.rknn --video_path video/test.mp4

REST API

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

Curl

curl http://localhost:8080/v1/chat/completions -d '{
  "model": "yolov6",
  "messages": [{"role": "user", "content": "Hello"}]
}'

Python

import requests

resp = requests.post(
    "http://localhost:8080/v1/chat/completions",
    json={"model": "yolov6", "messages": [{"role": "user", "content": "Hello"}]},
)
print(resp.json())

Model Details

reComputer RK

Armbian 26.8.7 Gnome or newer with OTA overlayroot: Configure Docker storage before running containers to avoid overlay mount errors.

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, dual-mode preview)

This deployment is based on the YOLOv6 modules in reComputer-RK-CV. It runs COCO object detection with RKNN-Toolkit-Lite2 and provides browser preview, REST inference, and offline video processing.

Model Information

Property	Value
Task	COCO 80-class object detection
Input	`640 x 640`, letterboxed and converted from BGR to RGB
Variants	`yolov6n.rknn`, `yolov6s.rknn`, `yolov6m.rknn`
Default	`model/yolov6n.rknn`
Output	Classes, confidence scores, and source-image boxes

Step A: Pull Images

bash

sudo docker pull ghcr.io/seeed-projects/recomputer-rk-cv/rk3588-yolov6:latest
sudo docker pull ghcr.io/seeed-projects/recomputer-rk-cv/rk3576-yolov6:latest

Step B: Run with One Click

Select the model size by changing --model_path.

For RK3576:

bash

sudo docker run --rm --privileged --net=host \
  --device /dev/dri/renderD128:/dev/dri/renderD128 \
  -v /proc/device-tree/compatible:/proc/device-tree/compatible:ro \
  ghcr.io/seeed-projects/recomputer-rk-cv/rk3576-yolov6:latest \
  python3 web_detection.py --platform rk3576 \
  --model_path model/yolov6n.rknn --video_path video/test.mp4

For RK3588:

bash

sudo docker run --rm --privileged --net=host \
  --device /dev/dri/renderD128:/dev/dri/renderD128 \
  -v /proc/device-tree/compatible:/proc/device-tree/compatible:ro \
  ghcr.io/seeed-projects/recomputer-rk-cv/rk3588-yolov6:latest \
  python3 web_detection.py --platform rk3588 \
  --model_path model/yolov6n.rknn --video_path video/test.mp4

Use model/yolov6s.rknn or model/yolov6m.rknn for the other catalog variants. Replace --video_path video/test.mp4 with --camera_id -1 for upload-only mode. Open http://<BOARD_IP>:8000 or /docs.

🔌 API Documentation

1. Model Inference Interface (Predict)

Endpoint: POST /api/models/yolov6/predict

Request Parameters (Multipart/Form-Data):

file: Optional image file to detect.
video: Optional MP4 file.
timestamp: Optional video timestamp in seconds; defaults to the first frame.
realtime: Optional boolean; with no upload, use the active camera or local-video frame.
conf: Optional request-level confidence threshold; defaults to 0.25.
iou: Optional request-level NMS IoU threshold; defaults to 0.45.

Usage Examples:

bash

curl -X POST "http://<BOARD_IP>:8000/api/models/yolov6/predict" \
  -F "file=@bus.jpg" -F "conf=0.25" -F "iou=0.45"

bash

curl -X POST "http://<BOARD_IP>:8000/api/models/yolov6/predict" \
  -F "video=@test.mp4" -F "timestamp=5.5"

Response Format (JSON):

json

{
  "success": true,
  "source": "uploaded image",
  "predictions": [
    {
      "class": "bus",
      "confidence": 0.91,
      "box": {"x1": 100, "y1": 120, "x2": 520, "y2": 430}
    }
  ],
  "image": {"width": 640, "height": 480}
}

2. System Configuration Interface (Config)

GET /api/config returns obj_thresh and nms_thresh. Update them with POST /api/config; request fields conf and iou override the global values.

The service also provides GET /api/health, GET /api/video_feed, POST /api/video/upload, POST /api/video/analyze, and video status, list, and download endpoints. Runtime files live under workspace/ unless RK_CV_WORKSPACE is set. Replacement models must retain the input size and YOLOv6 tensor layout expected by web_detection.py.

3. Command Line Arguments

Argument	Default	Description
`--platform`	Required	`rk3576` or `rk3588`; must match the image.
`--model_path`	Required	One of the packaged YOLOv6 RKNN files.
`--camera_id`	`1`	Camera index; `-1` enables upload-only mode.
`--video_path`	None	Looping local MP4; overrides the camera.
`--class_path`	COCO labels	Optional replacement class list.
`--host` / `--port`	`0.0.0.0` / `8000`	FastAPI listen address and port.

Real-time Video Stream Interface (Video Feed)

Get the latest annotated MJPEG stream for browser preview:

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

🛠️ Developer Guide (Production Recommendations)

Build Local Images

bash

docker build -f docker/rk3576/yolov6.dockerfile \
  -t rk3576-yolov6:local src/rk3576_yolov6

docker build -f docker/rk3588/yolov6.dockerfile \
  -t rk3588-yolov6:local src/rk3588_yolov6

Inputs and Outputs

Input: image, video frame, or camera frame. Output: COCO classes, confidence scores, and boxes.