I am trying to train my Convolutional Neural Network on full digital mammography images. Here are example dimensions from a random sample: (5832, 4104, 3). Which method for loading images and training the model is recommended in this case? I guess I would need to load only certain amount of images in memory so that it does not overflow. Should I use Keras ImageDataGenerator or TensorFlow's TFRecords?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.