Table 1 Key parameters used for fine-tuning the models. imgsz indicates the dimensions all images are resized to when training the model; epochs denotes the total number of complete passes over the full dataset for training; batch specifies the number of training images analyzed before updating the model’s weights; freeze determines the number of initial layers to be frozen with the weights remaining unchanged; amp enables automatic mixed precision; and half enables half precision inference.

From: Quantifying dislocation-type defects in post irradiation examination via transfer learning