🧬 Train a high-performance Yolov8 (CNN) model for histological analysis: data augmentation and combating overfitting by Active Learning

will be composed soon!

Briefly, I

• used a sliding window approach for data augmentation to handle high-resolution images or small objects
• initially trained YOLOV8 model shows overfitting (high variance): high performance on training data but low performance on testing data
• performed Active Learning by focusing on the “weakest” labels and adding more targeted data
• eventually, obtained a high-performance model