After Cleaning image do I need to train model again?

Jeremy first train model then clean data set , do I need to train model or not again?
In a book after cleaning he did not train again. Chapter 2 production

ImageClassifierCleaner shows you the images in training and validation datasets with high losses in descending order. With that, you can inspect images one by one and delete them if they don’t belong to any labels or move them to the correct ones.

cleaner = ImageClassifierCleaner(learn)
for idx in cleaner.delete(): cleaner.fns[idx].unlink()
for idx,cat in cleaner.change(): shutil.move(str(cleaner.fns[idx]), path/cat)

The above code deletes the images you want to delete and moves images to the correct label.

After that, you can train your model with the cleaned new datasets for better improvements.

3 Likes