search

hunglc007 / tensorflow-yolov4-tflite

YOLOv4, YOLOv4-tiny, YOLOv3, YOLOv3-tiny Implemented in Tensorflow 2.0, Android. Convert YOLO v4 .weights tensorflow, tensorrt and tflite
https://github.com/hunglc007/tensorflow-yolov4-tflite
MIT License
2.25k stars 1.24k forks source link

What is the purpose of custom BatchNormalization in common.py? #265

Open lostmsu opened 3 years ago

lostmsu commented 3 years ago

I am trying to figure out if it makes any sense for the training loop.

As far as I can see, freeze_all, that sets the trainable property, is only called on convolutional layers, so at least during training the new BatchNormalization has the same behavior as the original BatchNormalization from Keras, because its trainable is actually always True.

Can somebody clear my confusion here?