yolov4学习记录

YOLOv4采用的优化方法

Bag of Freebies (BoF) for backbone

数据增广手段，包括 CutMix and Mosaic data augmentation, DropBlock regularization,
Class label smoothing

数据增广在imgnet2012训练basenet

labelsmooth

正常情况下上面的公式会导致zi趋于无穷大，导致模型过度自信。因此修改后，zi不会趋于无穷大。经过我用mnist验证，可以明显提高百分之0.1的准确率。

Bag of Specials (BoS) for backbone

activation

Activations: ReLU, leaky-ReLU, parametric-ReLU,
ReLU6, SELU, Swish, or Mish
论文中采用Mish，实验验证的。

CSPNET

EfficientDet论文 (MiWRC)

优化结构

带权重的融合

Bag of Freebies (BoF) for detector

CIoU-loss,
CmBN, DropBlock regularization, Mosaic data aug-
mentation, Self-Adversarial Training, Eliminate grid
sensitivity, Using multiple anchors for a single ground
truth, Cosine annealing scheduler [52], Optimal hyper-
parameters, Random training shapes

GIoU, CIoU, DIoU

https://zhuanlan.zhihu.com/p/94799295

CmBN

当batch-size小时batch-normalization不准，因此通过利用之前得到的均值和方差来推理当前的均值。效果比较明显：

Additional blocks

SPP [25], ASPP [5], RFB[47], SAM [85]