YOLOv5 demo

YOLOv5 is a family of object detection architectures and models pretrained on the COCO dataset.

This demo for Ampere Altra/Max systems runs YOLOv5 object detection and classification ML inference using provided source video and outputs it to your browser at a smooth 30 frames per second. with bounding boxes showing classification confidence and inference latency in milliseconds. We may also add 60FPS examples.

Download the demo repository

$ git clone https://github.com/AmpereComputingAI/yolov5-demo.git
$ cd yolov5-demo

Install docker

Follow the instructions from this link, https://docs.docker.com/engine/install/ubuntu/

Install compose plugin

Follow the instructions from this link, https://docs.docker.com/compose/install/linux/#install-using-the-repository

Open the required ports

$ sudo firewall-cmd --zone public --permanent --add-port 5000-5010/tcp
$ sudo firewall-cmd --reload

Start the demo

There are two start scripts provided, one (start-yolo-cpu.sh) to run on CPU and the other one (start-yolo-gpu.sh) to run on GPU. Please select the video source (on the start script) and then run the script

$ ./start-yolo-cpu.sh
[+] Running 2/2
 ⠿ Network cont-1_default            Created                                               0.3s
 ⠿ Container cont-1-app-yolo-live-1  Started                                               0.5s

$ docker ps
CONTAINER ID   IMAGE                                            COMMAND           CREATED         STATUS         PORTS                                       NAMES
d220bd759eb5   ghcr.io/amperecomputingai/yolo-live-demo:1.2.7   "python app.py"   5 seconds ago   Up 4 seconds   0.0.0.0:5000->5000/tcp, :::5000->5000/tcp   cont-1-app-yolo-live-1

Open your favorite browser and type in the following URL.

http://<server-IP>:5000/app

Stop the demo

There are two stop scripts provided, one (stop-yolo-cpu.sh) to stop the CPU container and the other one (stop-yolo-gpu.sh) to stop the GPU container. Please use the appropriate one.

$ ./stop-yolo-cpu.sh

Select another video file as source

Edit start-yolo-cpu.sh (or start-yolo-gpu.sh) to select a different video source file by uncommenting it.

Reference

https://github.com/ultralytics/yolov5