In this case, the splitting is 80% for training, and 20% for validation is made. keras.backend.set_image_data_format('channels_first') # ...
確定! 回上一頁