YOLOv4 custom object detection with TensorFlow 2

To be a state-of-the-art model, YOLOv4 needs to be at the leading-edge of deep learning. YOLOv4 authors have worked on techniques to improve the accuracy of the model while training and in post-processing.

These techniques are called bag-of-freebies and bag-of-specials. In my previous tutorial, we mostly covered the YOLOv4 backbone, the neck and part of bag-of-specials (BoS) for the detector. So, in this part, we will continue bag-of-specials (BoS) and we'll cover bag-of-freebies (BoF).

✅ Support My Channel Through Patreon:

✅ One-Time Contribution Through PayPal:

[kavanpatelkvn]

You're really giving down to earth solution which we wanted. Thanks :)

[GarvChaudha]

Please make a video on explaining the yolo loss function
I can't understand why you use the raw yolo outputs for loss instead of decoded output

[Vladeeer]

hey man i been lurking around your channel for a while, what could you recommend watching to learn ML and in what order that will be relevant in today's world of TF, will be aiming study PPO once i get enough skill level

[zaheerbeg]

your mnist folder contains images in other format like dwg files?, you are not annotate images and its yolo format text files?

[thaongogia]

why the loss is nan when i train on GPU and seem nice when i train on CPU

[anggadharma]

is it possible to convert the model into tensorflow lite and quantized?

[lznnin]

can i use TPU to train my model using your code?

[밍슈슈]

I have some error like 'configure TEST options to validate model
You don't have saved model weights to measure mAP, check TRAIN_SAVE_BEST_ONLY and TRAIN_SAVE_CHECKPOINT lines in configs.py' this.
I use own my custom dataset and i changed configs.py, train.py follow your youtube.
But when i try python train.py, the same error occurs as before.

How can I solve this problem?

Thank u:)

[mustafagoksu]

How can I convert and use the weights we produce from Yolo models for this structure?

[zaheerbeg]

if i want to train[yolov4] model on custom images like mnist but in jpg format what change i have to do in -> files, like train images and yolo format annoted files files, kindly reply?

[MYance-utud]

Hey man! When i run in colab, into checkpoint doesnt create anything, some like yolov3 or 4_custom

[tingxiaoyang]

Hey, thanks for the channel, but I can not found the related YOLOV4 github codes. :d

[houston]

Text version tutorial link 404 not found