Skip to content

Files

Latest commit

vangjvangj
update labels
May 22, 2020
6117e84 · May 22, 2020

History

History

dl-darknet

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
backup
backup
Jul 31, 2019
cfg
cfg
Jul 31, 2019
data
data
Jul 31, 2019
image
image
Jul 31, 2019
log
log
Jul 31, 2019
scripts
scripts
Aug 1, 2019
video
video
Jul 31, 2019
.dockerignore
.dockerignore
Jul 31, 2019
.gitignore
.gitignore
Aug 1, 2019
DOCKERHUB.md
DOCKERHUB.md
Aug 13, 2019
Dockerfile
Dockerfile
May 22, 2020
README.md
README.md
Oct 14, 2019
build.sh
build.sh
Jul 31, 2019
deploy.sh
deploy.sh
Aug 13, 2019
setup.sh
setup.sh
Aug 1, 2019

README.md

One-Off Coder Logo

Purpose

The purpose of this docker image is to dockerize darknet so that you may easily use it and also in a portable manner.

Docker Hub

Image

Use as a command line

Detection with tiny weights.

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    -v $HOME/git/docker-containers/dl-darknet/cfg:/darknet/cfg \
    -v $HOME/git/docker-containers/dl-darknet/data:/darknet/data \
    -v $HOME/git/docker-containers/dl-darknet/image:/darknet/image \
    -v $HOME/git/docker-containers/dl-darknet/video:/darknet/video \
    -v $HOME/git/docker-containers/dl-darknet/log:/darknet/log \
    -v $HOME/git/docker-containers/dl-darknet/backup:/darknet/backup \
    -v $HOME/git/docker-containers/dl-darknet/scripts:/root/scripts \
    dl-darknet:local \
    detector test cfg/coco.data cfg/yolov3-tiny.cfg weight/yolov3-tiny.weights data/dog.jpg -dont_show > image/dog.log

Detection with normal weights.

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    -v $HOME/git/docker-containers/dl-darknet/cfg:/darknet/cfg \
    -v $HOME/git/docker-containers/dl-darknet/data:/darknet/data \
    -v $HOME/git/docker-containers/dl-darknet/image:/darknet/image \
    -v $HOME/git/docker-containers/dl-darknet/video:/darknet/video \
    -v $HOME/git/docker-containers/dl-darknet/log:/darknet/log \
    -v $HOME/git/docker-containers/dl-darknet/backup:/darknet/backup \
    -v $HOME/git/docker-containers/dl-darknet/scripts:/root/scripts \
    dl-darknet:local \
    detector test cfg/coco.data cfg/yolov3.cfg weight/yolov3.weights data/dog.jpg -dont_show > image/dog2.log

Detection on a MP4 file.

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    -v $HOME/git/docker-containers/dl-darknet/cfg:/darknet/cfg \
    -v $HOME/git/docker-containers/dl-darknet/data:/darknet/data \
    -v $HOME/git/docker-containers/dl-darknet/image:/darknet/image \
    -v $HOME/git/docker-containers/dl-darknet/video:/darknet/video \
    -v $HOME/git/docker-containers/dl-darknet/log:/darknet/log \
    -v $HOME/git/docker-containers/dl-darknet/backup:/darknet/backup \
    -v $HOME/git/docker-containers/dl-darknet/scripts:/root/scripts \
    dl-darknet:local \
    detector demo cfg/coco.data cfg/yolov3.cfg weight/yolov3.weights video/dummy.mp4 -out_filename video/dummy.avi -dont_show

Detection on a MP4 file and output to JSON + MJPEG + AVI. After you run the command below, direct your browsers to the following URLs.

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    -v $HOME/git/docker-containers/dl-darknet/cfg:/darknet/cfg \
    -v $HOME/git/docker-containers/dl-darknet/data:/darknet/data \
    -v $HOME/git/docker-containers/dl-darknet/image:/darknet/image \
    -v $HOME/git/docker-containers/dl-darknet/video:/darknet/video \
    -v $HOME/git/docker-containers/dl-darknet/log:/darknet/log \
    -v $HOME/git/docker-containers/dl-darknet/backup:/darknet/backup \
    -v $HOME/git/docker-containers/dl-darknet/scripts:/root/scripts \
    -p 8070:8070 \
    -p 8090:8090 \
    dl-darknet:local \
    detector demo cfg/coco.data cfg/yolov3.cfg weight/yolov3.weights video/dummy.mp4 -json_port 8070 -mjpeg_port 8090 -ext_output -dont_show -out_filename video/dummy.avi

Detection on a real-time video stream and redirect output to JSON + MJPEG + AVIG. Note that you can test the below by downloading and installing IP Webcam on your phone; replace the IP below with the one on your phone (the software on the phone will show you what the phone's IP is).

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    -v $HOME/git/docker-containers/dl-darknet/cfg:/darknet/cfg \
    -v $HOME/git/docker-containers/dl-darknet/data:/darknet/data \
    -v $HOME/git/docker-containers/dl-darknet/image:/darknet/image \
    -v $HOME/git/docker-containers/dl-darknet/video:/darknet/video \
    -v $HOME/git/docker-containers/dl-darknet/log:/darknet/log \
    -v $HOME/git/docker-containers/dl-darknet/backup:/darknet/backup \
    -v $HOME/git/docker-containers/dl-darknet/scripts:/root/scripts \
    -p 8070:8070 \
    -p 8090:8090 \
    dl-darknet:local \
    detector demo cfg/coco.data cfg/yolov3.cfg weight/yolov3.weights http://192.168.0.210:8080/video?dummy=param.mjpg -json_port 8070 -mjpeg_port 8090 -ext_output -dont_show -out_filename video/dummy.avi

Training your own object detector.

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    -v $HOME/git/docker-containers/dl-darknet/cfg:/darknet/cfg \
    -v $HOME/git/docker-containers/dl-darknet/data:/darknet/data \
    -v $HOME/git/docker-containers/dl-darknet/image:/darknet/image \
    -v $HOME/git/docker-containers/dl-darknet/video:/darknet/video \
    -v $HOME/git/docker-containers/dl-darknet/log:/darknet/log \
    -v $HOME/git/docker-containers/dl-darknet/backup:/darknet/backup \
    -v $HOME/git/docker-containers/dl-darknet/scripts:/root/scripts \
    dl-darknet:local \
    detector train /darknet/image/polygons/iaia-polygons.data /darknet/image/polygons/tiny-yolo-iaia-polygons.cfg -dont_show

Testing your own object detector.

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    -v $HOME/git/docker-containers/dl-darknet/cfg:/darknet/cfg \
    -v $HOME/git/docker-containers/dl-darknet/data:/darknet/data \
    -v $HOME/git/docker-containers/dl-darknet/image:/darknet/image \
    -v $HOME/git/docker-containers/dl-darknet/video:/darknet/video \
    -v $HOME/git/docker-containers/dl-darknet/log:/darknet/log \
    -v $HOME/git/docker-containers/dl-darknet/backup:/darknet/backup \
    -v $HOME/git/docker-containers/dl-darknet/scripts:/root/scripts \
    dl-darknet:local \
    detector test /darknet/image/polygons/iaia-polygons.data /darknet/image/polygons/tiny-yolo-iaia-polygons.cfg /darknet/backup/tiny-yolo-iaia-polygons_last.weights -ext_output -dont_show -out /darknet/log/result.json < /darknet/image/polygons/iaia-polygons_valid.txt

Annotating the images with the results.

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    -v $HOME/git/docker-containers/dl-darknet/cfg:/darknet/cfg \
    -v $HOME/git/docker-containers/dl-darknet/data:/darknet/data \
    -v $HOME/git/docker-containers/dl-darknet/image:/darknet/image \
    -v $HOME/git/docker-containers/dl-darknet/video:/darknet/video \
    -v $HOME/git/docker-containers/dl-darknet/log:/darknet/log \
    -v $HOME/git/docker-containers/dl-darknet/backup:/darknet/backup \
    -v $HOME/git/docker-containers/dl-darknet/scripts:/root/scripts \
    --entrypoint /opt/anaconda/bin/python \
    dl-darknet:local \
    /root/scripts/annotate.py -j /darknet/log/result.json  -d /darknet/image/polygons/annotations

Use in interactive terminal mode

To run local build with GPU support.

docker run -it \
    --runtime=nvidia \
    --shm-size=5g \
    -e NVIDIA_VISIBLE_DEVICES=0 \
    --entrypoint /bin/bash \
    dl-darknet:local

To test if OpenCV was installed correctly.

python -c "import cv2; print(cv2.__version__)"

To run Yolo v3 with tiny weights.

time ./darknet detect \
    cfg/yolov3-tiny.cfg \
    weight/yolov3-tiny.weights \
    data/dog.jpg

To run Yolo v3 with normal weights.

time ./darknet detect \
    cfg/yolov3.cfg \
    weight/yolov3.weights \
    data/dog.jpg

To run Yolo v3 full command.

time ./darknet detector \
    test cfg/coco.data \
    cfg/yolov3.cfg \
    weight/yolov3.weights \
    data/dog.jpg

Links

Take a Look!

Check out Grace Hopper.

Citation

@misc{oneoffcoder_dl_darknet_2019, 
title={Docker container with darknet}, 
url={https://github.com/oneoffcoder/docker-containers/tree/master/dl-darknet}, 
journal={GitHub},
author={One-Off Coder}, 
year={2019}, 
month={Jul}}