train_YoLo.py trains the YOLOv8 model for detecting chicken samples; detect_crop.py uses the trained target detection model to detect and crop the original image, while adding appropriate padding to maintain its proportion.
train.py is used to train deep learning models for chicken breed and gender classification, including 8 popular image classification models train_KFold.py divides the data into training and test sets using the K-fold crossover method, which is suitable for small data sets test.py calculates the commonly used evaluation index values for the trained deep learning model.
gradcam.py and swin_cam.py use Grad-CAM technology to generate the ROI of the model