Yolov4 trains its own data set

yolov4 yes yolo The fourth version in the series is improved from the previous three versions

yolov4 Run weighted residual connection (WRC), Cross phase partial connection (CSP), Standardization across small batches (CmBN), Self confrontation training (SAT),Mish Activate , Mosaic data enhancement ,CmBN,DropBlock Regularization ,CIoU Loss And other optimization methods to improve convolutional neural network (CNN) The accuracy of the .

This topic focuses on operation , Next, we will talk about how to use this yolov4 Algorithm to achieve their own data set training and detection .

Source code address ：

https://github.com/bubbliiiing/yolov4-pytorch

The code structure

  among ,

① img Stored are test pictures

② logs Stored training logs and training models

③ model_data Stored model configuration files and pre training models

④ nets Network related code

⑤ utils Other common interface files stored

⑥ utils_coco Stored for disposal coco Interface of format data

⑦ kmeans_for_anchors.py Computing data sets anchors

⑧ predict.py tr