I wonder how your build is triggered on push in the external repo?

These are the steps i have followed, im doing the setup on teslav100 ubantu 20.04 machine

1. docker pull daisukekobayashi/darknet:darknet_yolo_v4_pre-gpu
2. sudo docker run -it -d daisukekobayashi/darknet
3. docker exec -it da4112e1cf7b4d54aec7ca5c178abbb7284b3e69105ddb65248fbbcf5774abc5 bash
4. now im inside root directory of docker
5. $ git clone https://github.com/AlexeyAB/darknet.git
   $ cd darknet/build/darknet/x64
   $ curl -L -O https://pjreddie.com/media/files/yolov3.weights

6.docker run --runtime=nvidia --rm -v $PWD:/workspace -w /workspace daisukekobayashi/darknet darknet_yolo_v4_pre-gpu
darknet detector test data/coco.data yolov3.cfg yolov3.weights -i 0 -thresh 0.25 dog.jpg -ext_output

(unable to run it) Im facing issues with this command. not able to do anything

your help is really appriciate. Thanks in advance

i am trying to run darknet inside cuda docker container. my issue is that the cuda-devel base image which include the cuda toolkit has a size of over 3gb. i want to use the absolute bare minimum of cuda functionality in my docker image for running darknet in order to reduce docker image size.

any insight regarding this is appreciated.

Image detection works fine but video detection shows -

OpenCV isn't used - data augmentation will be slow 
Demo needs OpenCV for webcam images

even though I had opencv installed. I am using the darknet:cpu container.
Command:
docker run --rm -v $PWD:/workspace -w /workspace daisukekobayashi/darknet:cpu darknet detector demo data/obj.data cfg/yolov4-obj.cfg yolov4-cust.weights -i 0 video.mp4 -dont_show -out_filename results.avi

I am running the cpu-cv image, i doubt gpu will work on my machine as i have a very old radeon, inside the darknet folder as needed also with the yolov3.weights but any attempt at trying to run it will only give me some basic information and no detection. Could you please suggest what I am missing based on the below runs

Run detect

docker run --rm -v $PWD:/workspace -v /media/2TBStore/Media/Video/cctv/tp/2019/11/05:/images -w /workspace daisukekobayashi/darknet:cpu-cv darknet detect data/coco.data cfg/yolov3.cfg yolov3.weights  /images/Reolink401_TP_1_01_20191105081614.jpg -ext_output
 OpenCV version: 3.2.0
 GPU isn't used 
 Not used AVX 
 Not used FMA & AVX2 

Run detector test

docker run --rm -v $PWD:/workspace -v /media/2TBStore/Media/Video/cctv/tp/2019/11/05:/images -w /workspace daisukekobayashi/darknet:cpu-cv darknet detector test data/coco.data cfg/yolov3.cfg yolov3.weights  /images/Reolink401_TP_1_01_20191105081614.jpg -ext_output
 GPU isn't used 
 Not used AVX 
 Not used FMA & AVX2 
 OpenCV version: 3.2.0

Any hints?

I am running:

docker build --rm \
      --build-arg CONFIG="gpu-cv" \
      --build-arg ADDITIONAL_PACKAGES="libopencv-dev" \
      --build-arg SOURCE_BRANCH="master" \
      --build-arg SOURCE_COMMIT="a028bfa0da8eb96583c05d5bd00f4bfb9543f2da" \
      --tag test_yolo_darknet 

And getting "error: is not supported" when "configure.sh" is called.

Hi, thanks for your docker image, it helps me a lot with Yolo v4. I can run detecting on single image perfectly, but when I run to detect on multiple images and export to JSON file, seem like it cannot read the data in txt file.

docker run --runtime=nvidia --rm -v $PWD:/workspace -w /workspace daisukekobayashi/darknet:gpu \
	darknet detector test data/obj.data cfg/yolov4-obj.cfg yolov4-obj_final.weights -ext_output -dont_show -out result.json < data/test.txt 

I go to the code and try to log the input but it's null

CUDA-version: 11020 (11040), cuDNN: 8.1.1, GPU count: 1  
 OpenCV version: 4.2.0
 0 : compute_capability = 610, cudnn_half = 0, GPU: NVIDIA GeForce GTX 1050 
 Allocate additional workspace_size = 52.44 MB 
Loading weights from yolov4-obj_final.weights...start detecting:
net.optimized_memory = 0 
mini_batch = 1, batch = 16, time_steps = 1, train = 0 
Create CUDA-stream - 0 
 Create cudnn-handle 0 
nms_kind: greedynms (1), beta = 0.600000 
nms_kind: greedynms (1), beta = 0.600000 
nms_kind: greedynms (1), beta = 0.600000 
Done! Loaded 162 layers from weights-file 

 seen 64, trained: 384 K-images (6 Kilo-batches_64) 
Enter Image Path: input (null):

But when I run this command inside the container, it works successfully. Could you help me with this?

Would it be possible to have a script building for this combination of flags?

I'm very new in the world of containers and docker.

I'm planning to use docker to contain a qt applications that uses the darknet api on an linux enviroment, is darknet compiled with the lib option in this project? I need that the application works on windows, my problem is that the darknet api doesn't work very well on windows.

Would it be possible to call the include files from the darknet API from the docker to an windows environment?

Hi! I have tried setting up linux "motion" to show video in browser at a url something like http://localhost:8081. It displays / streams fine.

Problem is when I issue the command

docker run --runtime=nvidia --rm -v $PWD:/workspace -w /workspace daisukekobayashi/darknet:gpu darknet detector test data/yolo9000.cfg data/yolov3.weights data/9k.names data/9k.labels data/imagenet.shortnames.list http://localhost:8081 -c 0 -ext_output

it says
names: Using default 'data/names.list'

and then stops

not sure where to get names.list from

also how do i get it to work on the video stream and then output it do another http://localhost:8082 ?

Thanks in advance for all the help!

I adapt tag: gpu-cv-cc80 as yolo train image,
I fisrt trid use train command :

but it seems just display model summary, and not start training.

Secondly, I tried to compile darknet file by make command , but it threw out error:

Finally , I also try use detect command , no exception , it threw a error:

So I think it is due to missing cuda file, but I think doesn't it should be a completely darknet environment ? Could you give some idea to solve this problem

Hi,
I am getting the following error:

darknet: error while loading shared libraries: libcuda.so.1: cannot open shared object file: No such file or directory

My Dockerfile looks like:

FROM daisukekobayashi/darknet:gpu-cv-cc86
# I think the lines here are not relevant to the problem
...
darknet detect cfg/yolov3.cfg yolov3.weights data/dog.jpg

Thanks in advance,
Raza

I have precisely the same issue as reported here: NVIDIA/nvidia-container-toolkit#271
This seems to be fixed by a newer nvidia image, according to the referred issue.

Hi,

I'm trying to use, image daisukekobayashi/darknet:gpu-cv-cc75.
My GPU is: Quadro RTX 6000.
nvidia-smi is working inside the container showing cuda version 11.2.
nvcc --version inside the container is giving bash: nvcc: command not found.

When I'm trying to run darknet detector train/test... I'm getting the following error:

CUDA status Error: file: ./src/dark_cuda.c : () : line: 39 : build time: Jul 27 2021 - 19:08:59

CUDA Error: no CUDA-capable device is detected
Darknet error location: ./src/dark_cuda.c, check_error, line #70
CUDA Error: no CUDA-capable device is detected: Bad file descriptor

Any idea what am I doing wrong?
Thanks