Reference for the CNN: https://arxiv.org/abs/1703.08472
The file that creates the neural network, and trains it. Once the relevant paths are specified, will also save the variables, model, and other info.
The file that tests the neural network. Once the relevant paths are specified, will retrieve the trained network variables and other testing info.
Consists of all the line numbers in different files where you need to specify a path of your own choice. (What the path is used for has also been described)
Consists of all the dependencies that you need to install in order to run the network. This .yml file can be used to create a separate conda environment different from the 'base' conda environment. Use the folowing code: conda env create -f tfdl_env.yml
Consists of a sample Training class of images. To download the entire dataset, go to the following link: https://www.kaggle.com/moltean/fruits (Note:- I ignored the following classes in training and testing: Mangostan, Pear Kaiser, Tomato Maroon You can involve them too, but you will have to do some dataset-related modifications)
Contains required Dataset Related Modifications: (After removing the 3 classes mentioned above, there were 52,200 images in training set total. For each epoch, 200 of these images were used, picked at different indices 2 images each 522 spaces apart) Any dataset modification will require a number of changes in the code of cnn_saver.py.