search

pjreddie / darknet

Convolutional Neural Networks
http://pjreddie.com/darknet/
Other
25.82k stars 21.33k forks source link

Stuck on detection process start #2374

Open morpheus117 opened 3 years ago

morpheus117 commented 3 years ago

Hello all.

I am new to Darknet/YOLO. I have installed it in a Hyper-V Ubuntu VM properly and it works fine. Obviously the processing times are huge as I cannot passthrough the GPU as I have only one on my desktop PC.

So I have installed Darknet on Win10, follow all the painful procedure and it seems all is installed, but when I run any command as example:

.\darknet.exe detector test cfg/coco.data cfg/yolov3.cfg yolov3.weights -thresh 0.25

It got stuck for ages, 20-25 mins.

image

And when finally asks for the image it doesn't seems to identify the objects:

image

The environment:

The makefile looks like this:

GPU=1 CUDNN=1 CUDNN_HALF=1 OPENCV=1 AVX=1 OPENMP=1 LIBSO=0 ZED_CAMERA=0 ZED_CAMERA_v2_8=0

Another question (sorry about my lack of knowledge about Windows), the .\build.ps1 command is the one that makes use of the makefile hosted in Darknet folder? Because I have modified, even deleted the makefile and no difference when running the build.ps1 process.

Thanks all.

ecksate commented 3 years ago

I'm getting the same behavior in Ubuntu 20.04, RTX3070,CUDA 10.1,cuDNN 7.6. It works with GPU=0 (and it appropriately uses the CPU) but with GPU=1, it stops as shown in the screenshot.

popocry commented 3 years ago

I had the same problem, the loading time was too long, but I finally recognized the object. Windows10,Visual Studio 2019 Community,RTX3070,CUDA10.1,cuDNN 7.6

linksocarina commented 3 years ago

I also had the same problem, I installed darknet on Win10 and Ubuntu16.04, they both happened this problem. image

syed0596 commented 3 years ago

Same problem with yolov4. As mentioned above.

Screenshot_20210504-061301__01

syed0596 commented 3 years ago

Already answered here. https://github.com/AlexeyAB/darknet/issues/6853#issuecomment-714929167

Additionally CUDA 11.x is required with its corresponding cudnn version.

Hope this helps :)

DikshitV commented 2 years ago

I am facing the same issue. Request someone to guide me. image

DikshitV commented 2 years ago

It works perfectly fine when using google colab. Now, I am trying to run on ubuntu remote server with good GPU. The previous comments shows their model is failing during the testing phase. My code is failing during model training itself.