Set up a docker-container for object detection with darknet

using CUDA 10.2 on Ubuntu 18.04

Pre-launch setup

• get docker (https://www.docker.com/)

• get the nvidia-docker-toolkit (https://github.com/NVIDIA/nvidia-docker)

• add ppa with latest drivers:

sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt-get update

• get a nvidia display driver > 440:

sudo apt-get install nvidia-driver-440

• clone this repo and cd into it

• build the docker container (this may take a while)

docker build --build-arg user=${USER} -t darknet:latest .

• delete dangling docker images to free space

docker rmi $(docker images -f "dangling=true" -q)

Launch

• run the darknet.sh script to get into the container

sh ./darknet.sh

• follow https://pjreddie.com/darknet/yolo/ to use darknet

Settings

• change path to personal files at the end of the Dockerfile [Note: COPY paths are relative do the Dockerfile]

• change path to personal files in darknet.sh

• delete the --device entry in darknet.sh if you don't have a webcam attached