I am a bit confused. When and why I should use the following packages? For image data, should only the Keras image functions be sufficient?
a. Tensorflow 2.0 Dataset b. Python Pandas (or any other libs available for image and/or text data) c. Keras data generator, image augmentation etc.
Tensorflow dataset is similar to Pandas and Keras readers. However ts.dataset does not read the whole data to the memory, as in Pandas, instead it reads the data on the fly when needed. Also, tf.dataset is well optimized and more flexible for tensorflow. I also want to specify that tf.dataset have already had image generators, it is also possible that tf.dataset can work with Keras image genarators.