YOLOv3 Object Detection with Darknet for Windows/Linux | Install and Run with GPU and OPENCV

Learn how to get YOLOv3 up and running on your local machine with Darknet and how to compile it with GPU and OPENCV enabled! By the end of this video you will be able to run your own real-time YOLOv3 detections on images and videos!
#yolov3 #darknet #objectdetection
Install and Build OPENCV from my previous video: • How to Build and Insta...
Clone the code and learn more here: github.com/AlexeyAB/darknet
In this video I cover:
1. How to tell if you have a GPU on your machine.
2. How to properly install darknet for both Linux and Windows
3. Running YOLOv3 object detection on images and videos.
If you enjoyed the video, toss it a like! 👍
To Subscribe: kzbin.info/door/ryd...
Thanks so much for watching!
- The AI Guy

[@TheAIGuy]

Let me know if you have any issues or would like to see more YOLOv3 videos!

[@ogulcankarayel5625]

is it necessary setting opencv=1 ?

[@TheAIGuy]

@Oğulcan Karayel No! You can have opencv=0 and it will work without opencv!

[@ignacioswboada660]

Thanks for the video! In case you are looking for ideas, it would be great a video where we can know how many objects of each class were detected in the image. Foe example, if there are 2 zebras and 5 lions, a way to print on the cmd "Zebras = 2, Lions = 5".

[@TheAIGuy]

IgnacioSW Boada That sounds like a great idea. I will definitely work on that.

[@eustathiosandr8090]

Hello, i followed all your previous videos till this one, the only think didn't work just like yours its was in the CMake where i had to hit configure 2 times instead of 1 when i was installing OpenCV from Source. When i run the command on cmd for Darknet i came up with this error: "/darknet/darknet.exe: error while loading shared libraries: api-ms-win-crt-filesystem-l1-1-0.dll: cannot open shared object file: No such file or directory" I found and download the .dll file but nothing changed. Any idea what is wrong? Thanks in advance

[@sigmarulesog984]

Usually people save the day but you saved a month bro. Thanks.!

[@kida569]

You explained this really well, thank you!

[@x24870]

Great help for YOLO beginners, thanks man!

[@sarahch8878]

It's Finally work but without Cuda GPU ( I spent one week struggle with installing darknet, after that I set CUDA to 0) and Magic it works ! Thank you for this tuto !

[@creekw5246]

Thank you so much for the video!! It was very helpful!!

[@ogedaykhan9909]

Thanks a lot!!! After a week of struggle, I finally made it with your video. Hope best for your channel

[@TheAIGuy]

ismet okatar thanks! Glad you were able to get it running!

[@ogedaykhan9909]

@@TheAIGuy Hi Can I ask you something. I am trying to detect just oranges. and I realized that it is already trained in the model. How can I run the model for just detecting oranges but nothing else?

[@ogedaykhan9909]

@@TheAIGuy And can you advice any additional things in order to detect just oranges? Thanks in advance.

[@harshpatel8665]

I copy that file, still same error

[@davidecarone5465]

Hi, I have a question... At 15:07 in C/C++, Additional Include Directories you have included every folder "include" in opencv-contrib/modules/"everymodules"... as default I havent't already included them, do I have to include them all? Thanks

[@andresacosta576]

Thanks @The AI Guy for the video. I have a question: how can i use yolov3 for custom object using my web cam?

[@lowqchannel]

please do a guide about cmake. got error when building. "configuration incomplete"

[@surapornpuangpanbut]

Very good stuff, thank you ^^

[@jadalee4278]

how did you run the darknet.exe detect command? did you run it in the window's command prompt? I can't get mine to work

[@shawnlee2661]

if I install cuda and opencv those kinds of stuff in anaconda environment, what should I do to let my Cmake detect them

[@laxpors]

Hey Great tutorial! Just one question, I got everything to work, the only issue is that my image processing is saying it's taking 600 milliseconds to process images, even though i have the exact same GPU as you in the video (GTX 1650). I would like it to be as low as you show in the video, is there anything I can do to fix this?

[@GrzegorzKiernozek]

Awesome, thank you!

[@SirUGG]

GOD BLESS YOU BRO. You saved me

[@calitosmenem]

Thank you AI Guy, it helped me a lot. It worked fine for yolov4 too with minor changes. I had only one issue: pthreadVC2.dll missing. Copy darklib.dll and uselib.exe from darknet/Release folder to the darknet/ root folder and also copy pthreadGC2.dll and pthreadVC2.dll from darknet\3rdparty\pthreads\bin\ folder to the darknet/ root folder.

[@chocoboooruuuuu]

Thanks a lot dude

[@neviomarjanovic5581]

Hi, great explanation. Can I write my own logic for example a little program in python which will send image to darknet detection and count people in it ? Or is it just used to run like exe file.

[@johanneszwilling]

I had a CUDA NOT FOUND issue while building in cmake: Installing CUDA one more time (after Visual Studio) solved it. 🙃 Thank you for all the effort!

[@PatrickGovan]

do you have cuda, VS, and cmake all in the same drive?

[@johanneszwilling]

@@PatrickGovan As the Installation destination: Yes. Though I had the temp files for CUDA installation on a different drive

[@palanikumar9352]

I am having AMD Raedon r4 so, I cant use Cuda while building. Is it possible to use OpenCL instead to accelerate and make things faster?

[@sperberluis]

Hi! I have a problem when I try to run the cmd command C:\Repos\darknet\darknet.exe detect cfg/yolov3.cfg yolov3.weights data/dog.jpg -> CMD returns: CUDA-version: 10020 (10020), cuDNN: 7.6.5, GPU count: 1 OpenCV version: 4.3.0 Couldn't open file: cfg/coco.data I don't know why the detector doesn't work because cfg/coco.data file exists in my PC. Could you help me?

[@haojielim3623]

@The AI Guy, I followed the step and I found my GPU is not working properly as after configuring, there is no CMAKE_CUDA_COMPILER and also the CUDA and GPU are not automatically linked together, I checked that I have a GPU in laptop (piscal version)

[@sysadmin9396]

Can I just install dark net without cuda and open CV? For example if I wanted to use dark trace just to be able to create a data set that I created using the your lblImg video? I’ve watched your recent video with using yolo in a conda environment and got that working so I just need to know how to create a custom data set now without having to install opencv and cuda as I’ve been getting slot of problems :/. Thanks!!

[@lucasb6517]

for the nvidia driver, I think you can now just do 'conda install -c anaconda tensorflow-gpu' and it will automatically determine the necessary CUDA and cuDNN packages you need to install. Correct me if i'm wrong though, I don't want to give anyone bad advice

[@TheAIGuy]

Wow I'll definitely have to give this a try. Thanks!

[@joaovitordurcoferraz4599]

When i try to configure pops up that message CMake Error at CMakeLists.txt:202 (find_package): By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "OpenCV", but CMake did not find one. Could not find a package configuration file provided by "OpenCV" with any of the following names: OpenCVConfig.cmake opencv-config.cmake Add the installation prefix of "OpenCV" to CMAKE_PREFIX_PATH or set "OpenCV_DIR" to a directory containing one of the above files. If "OpenCV" provides a separate development package or SDK, be sure it has been installed. why i need to do, to fix that ?

[@aloufin]

My cuda architecture was set to "AUTO", and when generating got these warnings Policy CMP0104 is not set: CMAKE_CUDA_ARCHITECTURES now detected for NVCC, empty CUDA_ARCHITECTURES not allowed. ........... any idea what architecture to set in that field?

[@brobhai974]

have you found any solution ????

[@PeterMilko]

Does it take up a lot of GPU or can you play games with it running?

[@FarzanMNoori]

Hi I am getting this error D:\zotero\PhD\codes\8-darknet> ./darknet.exe detect cfg/yolov3.cfg yolov3.weights data/dog.jpg '.' is not recognized as an internal or external command, operable program or batch file. but unable to add variables

[@arturasdruteika2628]

Hi, nice tutorials. I have a problem, I'm using Linux, when I set GPU=1, CUDNN=1, CUDNN_HALF=1, OPENCV=1 in Makefile, and after running "./darknet -i 1 imagenet test cfg/alexnet.cfg alexnet.weights" I get an error: "GPU isn't used OpenCV isn't used - data augmentation will be slow Not an option: imagenet" Ihave cuda and opencv installed. Any suggestion on why it is not working?

[@adityakabu8190]

Hey, I am building it on Linux VM however in the last step that is .....imtest data.... command i get L2 Norm ;371.5189 something followed by Not compiled with OpenCV saving to original png instead. Could you help to resolve that?

[@naomirobinson5959]

did you solve this problem by any chance?

[@joiepatsy4991]

Excuse me sir,, i have 2 question. First, when i configured my darknet build in cmake it said my cuda compiler not found, and secondly, i wanna build without opencv because in my MakeFile i'd set OPENCV=0, but in my cmake also its said that OPENCV NOT FOUND, is that ok?

[@joiepatsy4991]

after the red block, i set enable opencv, and it is configured and generated, so far, its working properly even im not using opencv. thnkyou so much

[@lucatonelli4831]

My darkness.exe will run however the prediction image does not show the boxes and the command shell doesn't show any accuracies, please could you help me with this?

[@NGC-rr6vo]

does darknet.exe that we created use darknet_video.py code? can i modify this for my needs? how can i modify it? for instance i need to check traffic signs and do smth when sign was detected

[@kk12207e]

Can someone HELP!!! CMake Error at CMakeLists.txt:163 (find_package): By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "OpenCV", but CMake did not find one. Could not find a package configuration file provided by "OpenCV" with any of the following names: OpenCVConfig.cmake opencv-config.cmake Add the installation prefix of "OpenCV" to CMAKE_PREFIX_PATH or set "OpenCV_DIR" to a directory containing one of the above files. If "OpenCV" provides a separate development package or SDK, be sure it has been installed.

[@zainul2002]

hello, could you find the solution? I have the same issue right now

[@RS-vu5um]

Hi, Appreciate your effort in putting together these great videos on YOLOv3. Need guidance from you. Can you please let me know in what order it would be helpful to watch the Videos related to YOLO?

[@farjadhaider1034]

Hello! I am getting this error when I am building darknet: CMake Warning at CMakeLists.txt:103 (find_package): By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "OpenCV", but CMake did not find one. Could not find a package configuration file provided by "OpenCV" with any of the following names: OpenCVConfig.cmake opencv-config.cmake Add the installation prefix of "OpenCV" to CMAKE_PREFIX_PATH or set "OpenCV_DIR" to a directory containing one of the above files. If "OpenCV" provides a separate development package or SDK, be sure it has been installed. Any SOLUTION to this problem?

[@ROBYER1]

Did you find a solution?

[@rick28971]

Hello. What is MINGW64? I tried to google it but it doesnt have the same icon.

[@nimeshmakwana503]

Object being detected and image being saved, but opencv is not running, can you please help

[@adithya7461]

Hi can you help me with this getting this error I have nvidia Geforce 940MX and i installed CUDA 10.0 and while configure getting this error I gave the CUDA path too in CUDA_SDK_DIR as C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.0/ . even this getting this message CMake Warning at cmake/OpenCVFindLibsPerf.cmake:45 (message): OpenCV is not able to find/configure CUDA SDK (required by WITH_CUDA). CUDA support will be disabled in OpenCV build. To eliminate this warning remove WITH_CUDA=ON CMake configuration option. Call Stack (most recent call first): CMakeLists.txt:687 (include)

[@henslerkwan7794]

Hi, everything seem to run well except it doesnt run on GPU. Any idea how to fix this error?

[@daynialkhan718]

great work. and i enjoyed the tutorial. i am facing issues of "not an option: darknet" can you please help

[@haojielim3623]

I followed all the steps, but when I cmake the darknet file, I didn't see the CMAKE_CUDA_COMPILER and the list of items in the cmake is much lesser than you. HELP!!!

[@mohammedaminuumar5588]

Hey weldone for this great work. am trying to this using an Intel Iris Xe GPU or does it has to be NVIDIA GPU?

[@angelinakancheva3990]

Will the same steps work for Yolov2? Been trying to install Darknet forever now and reaching the point of desperation :( The installation is not trivial at all.

[@cemakgul2694]

Hey ! thanks for the nice video ! I have a problem with Cmake process. I was doing as u told and 11:59 on the video you got the no error but i have the error Could NOT find PThreads_windows (missing: PThreads_windows_LIBRARY PThreads_windows_INCLUDE_DIR) how can i fix it ? I checked online but not much of an info about it and im new to this. Thanks in advance !!

[@TheAIGuy]

in CMake make sure you have the check box in top right called "Advanced" checked. Then look for the variables in the CMake list that start with PThreads. Your PThreads_window_DIR should have value "C:/Repos/darknet/3rdparty/pthreads" and PThreads_window_INCLUDE_DIR should have value "C:/Repos/darknet/3rdparty/pthreads/include" . Try this but make sure to change your values to the right path for where you stored darknet as yours might not be C:/Repos

[@haojielim3623]

@@TheAIGuy Hi, I followed the step and I found my GPU is not working properly as after configuring, there is no CMAKE_CUDA_COMPILER and also the CUDA and GPU are not automatically linked together, I checked that I have a GPU in laptop (piscal version) Besides that, when I try to run the darknet.exe detect cfg/yolov3.cfg yolov3.weights data/dog.jpg, there is one error "the code execution cannot proceed because pthreadVC2.dll was not found."

[@taniadias9449]

I'm having some issues running the code. I have this error: Couldn't open file: cfg/coco.data. I'm not sure if is something missing in my instalations. Can you please help me?

[@zohaibjan]

I have double checked everything and followed the tutorial through and through but I keep gettings nans in my loss and it does not improve at all. Please advise

[@roydrori5100]

There Is no Cuda Download Version 10.0 Available for Win 11?

[@baraahamoood3372]

hi man i got this error can u help me with it : error while loading shared libraries: opencv_core460.dll: cannot open shared object file: No such file or directory

[@awdhootkanawade]

Facing a issue after last command shell doesn't display anything nor any prediction file , no errors either please help

[@Th3RobotWizard]

Hey, I know you made this video a while ago however when i try to launch darknet i get 4 errors 1 ) pthreadvc2 was not found 2) opencv.highgui440.dll was not found 3) opencv.imagecodecs440.dll was not found 4) opencv.videoio440.dll was not found itd be awesome if you responded and helped

[@johncaling6150]

check your paths

[@hepsibasolomon1755]

I got same error, I simply added these files among python installation location when I found python package don't have them. some of these files I already had from openCV downloads for other projects else directly download from git or google

[@mangaenfrancais934]

Hello, Thank you very much for your videos, I have an NVIDIA GeForce MX130, and if I am not wrong, I find that I have to use CUDA SDK 5.0, but I am really confused the date of the release of 5.0 is (Oct 2012) and the release of NVIDIA GeForce MX130 is Nov 17th, 2017, do I have to use the version 5, or I can use the latest version, I need your help and thank you.

[@keshavkantsharma1946]

Hey, you can download the latest SDK 10.2, as it supports Maxwell microarchitecture. So u r good to go ahead with the latest SDK. MX130 has Maxwell

[@BHARDWAJ1991]

Will this work if I do not have CUDA. A/c to Wikipedia I should be using CUDA 8.0 toolkit. But that is not available for my OS - Ubuntu 18.04

[@harshchoubey8440]

CMake Error at CMakeLists.txt:202 (find_package): By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "OpenCV", but CMake did not find one. Could not find a package configuration file provided by "OpenCV" with any of the following names: OpenCVConfig.cmake opencv-config.cmake Add the installation prefix of "OpenCV" to CMAKE_PREFIX_PATH or set "OpenCV_DIR" to a directory containing one of the above files. If "OpenCV" provides a separate development package or SDK, be sure it has been installed. Can somebody help me with this error?

[@sarahch8878]

Hi, did anyone have a problem with "Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.28.29910/bin/Hostx64/x64/cl.exe - skipped" skipped and not works ? I spent a lot of time installing packages and I still have errors in Cmake to build the darknet !

[@Kryptosels11]

please let me know how to set the policy in 3.18.2 ...i cant generate sln because of this issue

[@siminesfshani8668]

I get pthreadvc2.dll not found when doing the above process. How I can solve this? I really appreciate your help.

[@amanineffati6546]

hello @The AI Guy i have nvidia geforce gtx 1050 and windows 10 i've installed cuda 8.0 and I've followed your previous tutorial and installed OpenCV on my computer. But when I configured the darknet code with CMake, I got a warning: CMake Warning at CMakeLists.txt:96 (find_package): By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "OpenCV", but CMake did not find one. Could not find a package configuration file provided by "OpenCV" with any of the following names: OpenCVConfig.cmake opencv-config.cmake Add the installation prefix of "OpenCV" to CMAKE_PREFIX_PATH or set "OpenCV_DIR" to a directory containing one of the above files. If "OpenCV" provides a separate development package or SDK, be sure it has been installed.

[@ROBYER1]

Did you have a fix for this?

[@aqiff123]

I have an error to configure this on windows. "Error in configuration process, files may be invalid". I had configured CUDA and CUDNN since last year and even use it. So can I know what's wrong ?

[@takitazwarparthib3555]

Same problem here :/

[@murtazajabalpurwala8124]

I am facing the same error

[@teetanrobotics5363]

Could you please order the playlists chronologically,so it becomes easier to follow the videos in a sequence ? Thanks for amazing content.Have watched almost all your videos

[@ruksharalam173]

I agree

[@hani97tn]

thanks for this video , it works but the problem is when i excute the command to predict he tells me GPU isn' t used i dont knowwhy i have cuda 9 cudann 7

[@bobopidgey5724]

i have the same problem, any solution?

[@aakashgupta1486]

i had followed every step you told while checking opencv it says ModuleNotFound Error: No module named 'cv2'. ms visual studio had seen no errors then why this happen

[@TheAIGuy]

Akash Gupta you must follow one of my previous videos “How to Build and Install Opencv from Source” in order to get cv2. Have you tried this?

[@blackunicorn_tm]

My cmake gui said vcpkg not found even though i’ve installed it. How do i solve the problem?

[@ternjiachie9167]

i have follow all the step, but why after prediction, there is no inference time in the result?

[@phongvo7641]

Hi I was attempting to build the ALL_BUILD in the Darknet.sln file and the output prompted me with 2 successed and 4 failed. Briefly scrolling through the error list I noticed dozens of lines referring to syntax errors. "Expected an identifier", "Expected a ;", "Expected an explicit type (int assume), etc. A full line looks like this, "2>C:\MinGW\include\wchar.h(435): error : expected an identifier"

[@phongvo7641]

I am working with your YoloV3 in the cloud but I would much prefer to be able to install this on my local machine. I have CUDA and cuDNN installed. Everything worked fine until the build in the darknet vs project.

[@phongvo7641]

I fixed my problem, uninstalled mingw from my computer, wiped the darknet clone and started this video from scratch. My new "dilemma" is that when I build the darknet ALL_BUILD I get 5 succeeded and 0 failed, not 6 succeeded.

[@fatimaslim6838]

error while loading shared libraries: api-ms-win-crt-locale-l1-1-0.dll: cannot open shared object file: No such file or directory I face this problem when i run MINGW64 please help me

[@charlesedison9638]

Does it work if I just download the binary opencv instead of building it from source? it takes hours to build it

[@TheAIGuy]

yes i believe that could possibly work. You can also check out my "Yolov3 in the Cloud" video, it will show you how to install and run yolov3 with darknet in minutes!

[@chittineediteja7330]

Configuring incomplete, errors occurred! See also "F:/darknet/build/CMakeFiles/CMakeOutput.log".

[@koritorres09]

Did you solved this problem?

[@husseingharakhani8002]

Thank you so much for your amazing work. I need to train my own model in order to detect cotton bolls. So I would be more than pleased if you make a tutorial for that or give a hint. Also, I am working on Jetson TX2 and my speed for live camera was about 2FPS and for one image the time was about 1800 ms. Is this speed normal for Jetson TX2 kit?

[@Acemarcelo12]

He has video on how to do that

[@vitoriagarcia9876]

If my 'device manager' doesn't show a GPU, how can I proceed?

[@schastain007]

When I run ./darknet.exe detect cfg/yolov3.cfg yolov3.weights data/dog.jpg I get this error error while loading shared libraries: api-ms-win-crt-locale-l1-1-0.dll: cannot open shared object file: No such file or directory I got no errors when running OpenCV or CMake not sure what this file is or how to fix it

[@josephblessingh2384]

Were you able to fix it? Was getting the same error!! But I fixed it. First of all make sure you are running the right command: ./darknet.exe detect cfg/yolov3.cfg yolov3.weights data/dog.jpg If still not working then, Follow this answer. stackoverflow.com/a/43305804/9581308 Check if you have ucrt under Windows Kits\10\ or not. If not, then open visual studio -> Tools -> Get Tools and features. Under that go to Individual components and search for crt sdk. Install it. Set the System Variable Path according to the stackoverflow answer. Just add 2 new entries to the System Path and you will be googd to go. Once you run the command, you will probably get pthread error. If yes, then open up 3rdparty folder -> pthreads -> bin Copy paste the 2 dll files to darknet folder. Run the command and voila. Detection should happen!! Hope I helped you, if still facing error then reply.

[@TheAIGuy]

try adding this to your PATH system variable, "C:\Windows\System32\downlevel" and then restart your computer and try again!

[@KeunFish]

How can we save detections in json format in firebase realtime ?

[@ronaktawde]

I have graphics card which i quite old version 3.2 as per search done on wiki. so should i install the cuda or not

[@mohammedammaruddin3781]

no

[@anirbansarkar6306]

I downloaded the darknet from github, but I don't see any 'build' folder in it

[@nadaabbas6329]

is the weights file downloaded compatible for YOLO Tiny? I can't find a similar weights file for YOLO Tiny

[@TheAIGuy]

nada abbas you can download tiny yolo weights from the official website by pjreddie

[@nadaabbas6329]

@@TheAIGuy thanks

[@user-ic7tn7fx1d]

There are no cuda.libs in OpenCV\build\install\x64\vc16\lib. I checked build -> cuda.lib on CMake. But it can't be created. Can you tell me why? Using opencv4.2 / cuda10.2 / python 3.8.

[@TheAIGuy]

이종선 did you build opencv with the with_cuda flag enabled?

[@user-ic7tn7fx1d]

@@TheAIGuy The sequence of progress is as follows. The first did not check with_cuda. And ran Visual Studio and ran openv build. The second time I checked with couda. And ran Configure and Generate in CMake. (opencv build was not.) If you check with_cuda and run openv build, there are many warning errors.

[@TheAIGuy]

@@user-ic7tn7fx1d you could try using my yolov3 on the cloud video as it wont have warnings like these

[@meudta293]

cant enter the Cmake website to donwload the tool can someone give me a link or something

[@ravirajsinhdabhi9255]

instead of showing bounding boxes in images and video, how to extract feachers like detected classes in images and coordinates of detected class???

[@TheAIGuy]

Ravirajsinh Dabhi check out my “How to Build Object Detection APIs” video where i actually already do this!

[@hasaniqbal1646]

12:01, I do all your steps but I get NO CMAKE CUDA COMPILER FOUND. When i set the value of CMAKE_CUDA_COMPILER = C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.2/bin/nvcc.exe, i get error NO CUDA TOOLKIT FOUND. Help?

[@jshr3]

Same here :( Could you solve it?

[@angelomoro8430]

Hi! thanks for the video! I'm running now darknet on cmd (win10) and cannot find these pthreadVC2.dll, opencv_highgui420.dll, opencv_videoio420.dll, opencv_imgcodecs420.dll. Do you maybe know why?

[@TheAIGuy]

copy the dll file from darknet/3rdparty/pthreads into the darknet folder and that should fix it! Look in your opencv install for the other two and move them also to darknet folder.

[@keshavkantsharma1946]

It solved! thanks, AI GUY!

[@papabudzz]

@@TheAIGuy ​ @Keshavkant Sharma already moved pthreadVC2.dll, opencv_highgui420.dll, opencv_videoio420.dll, and opencv_imgcodecs420.dll into the darknet folder but darknet.exe app isn't opening. Any help?

[@yazuroichi7687]

@@TheAIGuy moved the dll files but still no luck, the darknet.exe app isn't opening.. what should i do?

[@murtazajabalpurwala8124]

The cmake couldnt compile. The process file is invalid. Can you please help

[@ruksharalam173]

I have built alexey's darknet in my windows 10 using VC 2019/version 16. but running the darknet.exe file, i face errors saying several opencv dll are missing despite configuring the path in VC. Has anyone else faced such problems? Any suggestion is welcome.

[@zainul2002]

hello, could you fix the issue?

[@Acemarcelo12]

Hi, My laptop does not have GPU(nvidia GetForce GTX). can this also work for me?

[@TheAIGuy]

azeez akeem i would recommend following my videos on how to train a custom yolov3 or yolov4 model in the cloud for free. Will be much faster for you since you don’t have nvidia GPU!

[@sumanabasu3339]

Hi, I have install CUDA, OpenCV successfully. Build the darknet as well. However when I am trying to run detection through command line, it is not working. It not even starts to run. Please help. Thank you

[@PatrickGovan]

Hi Sumana, I am having the same issue. were you able to resolve this?

[@sailfromsurigao]

I have this error when trying to follow your video, at the part where you run cmake in darknet. Please help. CMake Error at CMakeLists.txt:163 (find_package): By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "OpenCV", but CMake did not find one. Could not find a package configuration file provided by "OpenCV" with any of the following names: OpenCVConfig.cmake opencv-config.cmake Add the installation prefix of "OpenCV" to CMAKE_PREFIX_PATH or set "OpenCV_DIR" to a directory containing one of the above files. If "OpenCV" provides a separate development package or SDK, be sure it has been installed.

[@dhruvsantoshwar6080]

Looking for a CUDA compiler - NOTFOUND Looking for pthread.h - not found CMake Warning at CMakeLists.txt:103 (find_package): By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "OpenCV", but CMake did not find one.

[@techbuildspcs]

Did you solve the issue?

[@parthasarathigoswami1982]

when I'm configuring the OpenCV from CMake with WITH_CUDA its successfully configuring and generating the build files there. But with the same when I'm going for OpenCV.sln in VS then its giving me lots of errors in ALL_BUILD - build. But at the same time if I ignore WITH_CUDA in CMake then its successful in MS VS build. Can you please help me?

[@TheAIGuy]

What are your errors like? Also if you check out my new video you can do Yolov3 in the cloud without having to build opencv or darknet!

[@parthasarathigoswami1982]

@@TheAIGuy lots of errors .. something like for different lib files: LINK : fatal error LNK1181: cannot open input file '..\..\lib\Release\opencv_highgui412.lib'

[@TheAIGuy]

saptech engineer I would use my latest video on doing it in the cloud! Its super easy, only a few commands.

[@pkz111111]

@@TheAIGuy Same issue when building OpenCV 4.4.0 from source. When I build "ALL_BUILD" and "INSTALL" projects using VS2019 Windows10 I got 2000 somthing warnings (only warnings no errors at all) saying: "Warning: field of class type without a DLL interface used in a class with a DLL interface" and "Warning: base class dllexport/dllimport specification differs from that of the derived class" Pre install CUDA 10.1 with matching cudnn. GPU: RTX 2080. When tring to build darknet for using YoloV3 it dosen't showing CUDA in cMake.

[@rifatislam5836]

Bro does this code give confusion matrix?

[@artyomharutyunyan5777]

I have problem during configure in CMake. Error ''cant found any instance of visual studio'' But I have already installed VS. How can I fix this problem ?

[@aleemrehmtulla]

there is a diffrence between vscode and visual studio community

[@roble.lawrence15]

I don't have GPU, is there alternative for this?

[@JNK369]

Hello AI guys , i hace have no GPU what can i do , is there any essential to download a CUDA?

[@TheAIGuy]

if you dont have a GPU then you dont need to download or install cuda, just leave the "WITH_CUDA" box unchecked. You should check out my "Yolov3 in the cloud..." video to see how you can train custom model with a free GPU in the cloud. Will speed it up a lot for you.

[@davidecarone9182]

Hi, when I tick on ENABLE_CUDA and I run "Configure" cmake say "Error in configuration process, project files may be invalid".. I have compared each row of cmake with yours and they are the same

[@TheAIGuy]

Are you sure you have CUDA downloaded and configured properly?

[@davidecarone9182]

@@TheAIGuy Yes, I have followed every step on this video. I have cuda on environment variable too

[@TheAIGuy]

@@davidecarone9182 Try running a cuda sample program to make sure that your cuda is running properly. Can find them here: docs.nvidia.com/cuda/cuda-samples/index.html#getting-started-with-cuda-samples

[@hecate3360]

@@TheAIGuy Have the same problem with "No CUDA toolset found"

[@christianjreyes21]

@@TheAIGuy I am also having the same problem. Has anyone found a fix for this yet?

[@tanujaa4683]

I am not able to get bounding box on prediction image

[@johncaling6150]

I get Could not find toolchain file: druing configuring.

[@seeman4171]

I have Radeon (TM) 520 graphics, can darknet work for it. please help me sir

[@TheAIGuy]

I would check out my latest video on how to use yolov3 in the cloud. It will let you use darknet guaranteed.

[@Pkm955]

Hi, I'm not sure if you will reply my comment, but I am facing a problem, when I configure darknet, it has this error CMake Warning at CMakeLists.txt:103 (find_package): By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "OpenCV", but CMake did not find one. Could not find a package configuration file provided by "OpenCV" with any of the following names: OpenCVConfig.cmake opencv-config.cmake Add the installation prefix of "OpenCV" to CMAKE_PREFIX_PATH or set "OpenCV_DIR" to a directory containing one of the above files. If "OpenCV" provides a separate development package or SDK, be sure it has been installed. I checked and those files exists but it is not finding them, could you help me on this?

[@rileykim13]

I have the same problem :/

[@ROBYER1]

@@rileykim13 Any fix?

[@joiepatsy4991]

can i use opencv-python and not using opencv built from source?

[@TheAIGuy]

Joie Patsy you can not use opencv-python because the darknet code is in C

[@keshavkantsharma1946]

Hello, Thanks for the wonderful video, I am facing an error while running my recorded video(48MB) using the command as u taught. I saved the video in data folder. "OpenCV isn't used - data increase will run slowly demo needs OpenCV for webcam images." I did change the settings in Makefile by changing the binary of openCv, openCV = 1. But still i am facing the same error. Please help me out. Thanks in advance.

[@davidecarone5465]

i have the same problem, any solution?

[@sarthakbhole3724]

Change it manually in Cmake provide path to opencv. There's a comment at top explaining this

[@aidanazkafaro5601]

Is it possible to use this approach without a GPU?

[@parikshitbangde7599]

C:/Users/admin/darknet/darknet-master/darknet.exe: error while loading shared libraries: pthreadVC2.dll: cannot open shared object file: No such file or directory what to do with this error Thanks

[@TheAIGuy]

Parikshit Bangde copy the dll file from darknet/3rdparty/pthreads into the darknet folder and that should fix it!

[@peregrinussolutionsllc6010]

@@TheAIGuy This same error popped up for me as well, but your comment resolved the error! :-)

[@peregrinussolutionsllc6010]

I have Cuda and successfully installed OpenCV from source by following your other video, and I enabled/check the correct boxes earlier in this video with CMake, but I receive this message at the very end after running the C:/Users/admin/darknet/darknet-master/darknet.exe: "Not compiled with OpenCV, saving to predictions.png instead" Should I try Configuring and Generating in CMake again before trying in the command prompt? The predictions image doesn't pop up automatically, but it is there. I tried to run it with the webcam, but it says it requires OpenCV to be enabled.

[@aikoyuuki9291]

@@peregrinussolutionsllc6010 Same problem, if you get a hint, let me know sir