This document discusses using neural networks for image compression. It describes traditional image compression techniques and architectures, as well as different types of neural networks like convolutional neural networks (CNNs) and recurrent neural networks (RNNs). CNNs have been used in autoencoders to obtain a latent space representation of an image for compression. RNNs have also been proposed for neural network image coding frameworks to leverage their memory capabilities. Examples show neural networks can achieve comparable or better compression than traditional techniques like JPEG at lower bitrates.