Train a custom object detector using YOLOv4 and YOLOv4-tiny (on your WINDOWS system)

👉🏻This video shows step by step tutorial on how to train a custom YOLOv4 object detector using darknet on your local WINDOWS system. In this tutorial, I have trained a custom YOLOv4 detector for mask detection.

① ⚡⚡ My Website Blog post on YOLOv4 & YOLOv4-tiny ⚡⚡

② ⚡⚡ Medium post on YOLOv4 & YOLOv4-tiny ⚡⚡

③ ⚡⚡ Video on How to Install and run YOLOv4-Darknet on Windows ⚡⚡

④ ⚡⚡ My GitHub link for Windows custom YOLOv4 training files ⚡⚡

⑤ ⚡⚡ My GitHub link for Windows custom YOLOv4-tiny training files ⚡⚡

⑥ ⚡⚡ Github link for Official Darknet repository ⚡⚡

⑧ ⚡⚡ Pre-trained yolov4 weights file ⚡⚡

⑨ ⚡⚡ Pre-trained yolov4-tiny weights file ⚡⚡

⑩ ⚡⚡YOLOv4 vs YOLOv4-tiny - A comparison ⚡⚡

#yolov4 #objectdetection #yolov4onwindows
#yolov4-tiny

▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬

▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Making these videos takes a lot of time and effort, so if you like these videos and if you can, then please support the channel using any of the following:

► Buy me a coffee! ☕

► Support channel on Patreon! 🖖


▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Thanks for watching!

#yolov4tiny #customyolov4 #customyolov4tiny #customobjectdetector #objectdetector #yoloobjectdetector #traincustomyolov4 #trainyolov4 #maskdetection

[ujmvvdw]

Thank you for your sharing.
This video is help me a lot.

[keerthanayelchuri]

I did everything mentioned in the video. I got all the weight files successfully too. But when I try to test an image, no object is being detected. Any idea why that might happen? I know there is no issue with data set for sure, and I also changed the yolov4-custom.cfg file before testing.

[lznnin]

cuDNN status Error in: file: function: line: 553

cuDNN Error: CUDNN_STATUS_BAD_PARAM
Darknet error location: C:\Users\cw_99\Downloads\darknet-master\src\convolutional_kernels.cu, forward_convolutional_layer_gpu(), line #553
cuDNN Error: CUDNN_STATUS_BAD_PARAM: No error

i got this error after 1000 iterations? Could you please help me?

[kfirdaus]

Hi @techzizou. Do you know how to do prediction on an annotated image? Bcos i want to compare the prediction with ground truth and see the prediction performance. Anyway, thanks for this detail tutorial. Really appreciate your effort!

[utkarshshivhare]

Hi, I am getting an error
Could not locate zlibwapi.dll. Please make sure it is in your library path!
could you please help me with that

[darkloneblade]

Does anyone know who to extract the Inference Time from the model performance check?

[HeyclipS]

Hello, I followed the whole tutorial and when I compile with the darknet.exe detector train data/obj.data cfg/yolov4-custom.cfg yolov4.conv.137 -dont_show -map, it perfectly works

However, I only have yolov4-custom_last at the end, I saw on google that is due to detector.c fault, can you give send me your file please ?

[liencnpa]

Hello ~
I have 2 classes.
When I change change filters=21 at cfg file before each [yolo] layer as your guide "keep in mind that it only has to be the last [convolutional] before each of the [yolo] layers". Then Run detector on an image, no object is being detected.

But I change all line in cfg file, all filters set to 21 (filters=21), Then Run detector on an image, object is being detected but only one classes detected.

I don't know why ? Can you explain more ?

Thank you !

[danaali]

HI, I have this error(
CUDA Error: out of memory) what should i do to fix this error please help me

[andrepinto]

I did everything mentioned in the video. I got all the weight files successfully too, treined with 4.000 images. But when I try to test an image, no object is being detected. Any idea why that might happen? I know there is no issue with data set for sure, and I also changed the yolov4-custom.cfg file before testing.

[joelbhaskarnadar]

Can you explain why the bounding boxes is not perfect for some images

[standpunkt_reddit]

I am getting this error changed everything according to the video
Error: l.outputs == params.inputs
filters= in the [convolutional]-layer doesn't correspond to classes= or mask= in [yolo]-layer

[raflisyawal]

sir please help me, i got error like this "Error: cuDNN isn't found FWD algo for convolution.", what should i do? thank you very much

[RajSharma-uivq]

How can I use the output of YOLO to another program. Suppose, to display the text if the person has not put on the mask? Any idea?

[RajSharma-uivq]

I did exactly the way you mentioned. It worked and the training is completed. However, I want to increase the number of Epochs now. So, how can I do that? Since, the training is already completed, can I increase the epochs and resume the training? Thank you so much for this wonderful video.

[alexaelena]

Hello! I have a problem with Command Prompt. When I'm typing darknet.exe detector train data/obj.data cfg/yolov4-custom.cfg yolov4.conv.137 -dont_show -map for training, it show me the error "couldn't open file : data/obj.data". What can i do to fix it?

[RajSharma-uivq]

Great video. BTW I encountered an error. "Wrong annotation: class_id = 1. But class_id should be [from 0 to 0]; file.. " I tried using only one class. So what changes should I make? Thank You.

[lznnin]

Is it possible to use tpu instead of gpu in your code? or is there something I need to change?

[kareemessam]

I have a question bro
I trained the model till it finishes
Now i added new images for trainning and want to continue trainning not beginning from scratch
when i use this command: darknet.exe detector train data/obj.data cfg/yolov4-custom.cfg -dont_show -map
it give me that:
Loading weights from
seen 64, trained: 128 K-images (2 Kilo-batches_64)
Done! Loaded 162 layers from weights-file
Learning Rate: 0.001, Momentum: 0.949, Decay: 0.0005
Detection layer: 139 - type = 28
Detection layer: 150 - type = 28
Detection layer: 161 - type = 28
Saving weights to
Create 6 permanent cpu-threads
If you want to train from the beginning, then use flag in the end of training command: -clear

[ravinpech]

Hello! I have difference labels txt from yours, my label just like this [Football 121 1212 12122 2121]. How can I change it?