Deploy a object detection projects rapidly in just 3 steps

When we usually work on personal demo designs, does our lack of professional expertise become an obstacle when deploying algorithms like object detection or pose estimation, preventing us from realizing our ideas? Or, as professional developers, do we still have to go through the tedious process of downloading models, configuring environments, and writing business code over and over again? Or are we overwhelmed by the complex inference code of various platform examples?

After testing multiple platforms recently, I have now summarized the fastest project deployment method. It does not require professional technical skills or an in-depth understanding of the domain, nor does it involve cumbersome steps. With just 3 steps, you can freely and rapidly deploy various models—CV, LLM, VLLM, SOUND—and accelerate development!

This method has 3 steps in total.

Step 1 – Visit the reComputer AI Lab platform

In any search tool, search for and visit the reComputer AI Lab platform, then enter the model interface: Models | reComputer AI Lab

Step 2 – Get the download command

On the model platform, we can select the model we need based on our requirements. Here we take the pose detection model, the YOLOv8n‑pose model, as an example. After selecting the model, click "Run".

In the pop-up window, select the device model you own. Currently, I am using the reComputer RK3576 device. In addition to this device, there are also other models such as RK3588, CM5+Hailo8/10, Jetson, and more. After that, obtain the download command for the original case.

sudo docker run --rm \
--name rk3576-yolov8n-pose \
--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

Step 3 – Log in to the device and run the case

Remotely connect to your personal device and download the Docker image (if already downloaded, you can skip this step).

curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh --mirror Aliyun
sudo systemctl enable docker
sudo systemctl start docker

After completing the Docker download, directly execute the command you copied above in the terminal. Upon successful execution, you will obtain a web URL.

After clicking to navigate to the corresponding interface, you will see the successful inference screen.

Running other cases is just the same:

Object Detection:

sudo docker run --rm \
--name rk3588-yolov8n-object \
--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/rk3588-yolov8:latest \
python3 web_detection.py --model_path model/yolov8n.rknn --video video/test.mp4

LLM：

sudo docker run --rm -it \
--name rk3576-1-5b-fp16 \
--privileged \
--net=host \
--device /dev/dri \
--device /dev/dma_heap \
--device /dev/rknpu \
--device /dev/mali0 \
-v /dev:/dev \
ghcr.io/seeed-projects/rk3576-deepseek-r1-distill-qwen:1.5b-fp16-latest

VLM：

sudo docker run --rm -it \
--name rk3576-3b-w4a16 \
--privileged \
--net=host \
--device /dev/dri \
--device /dev/dma_heap \
--device /dev/rknpu \
--device /dev/mali0 \
-v /dev:/dev \
ghcr.io/seeed-projects/rk3576-qwen2.5-vl:3b-w4a16-latest

SOUND：

sudo docker run --rm \
--name rk3576-whisper \
--privileged \
--net=host \
-e PYTHONUNBUFFERED=1 \
-e RKNN_LOG_LEVEL=0 \
-v /proc/device-tree/compatible:/proc/device-tree/compatible \
ghcr.io/seeed-projects/recomputer-rk-cv/rk3576-whisper:latest \
python3 web_service.py