A simple image autoencoder for the CIFAR 10 dataset. Compresses images by over 80% then learns to reconstruct them. CIFAR images are 32x32 pixels, which is why the original images are such low quality. This is just a proof of concept until I can obtain more computing power.
wjegbert / autoencoder Goto Github PK
View Code? Open in Web Editor NEWA simple convolutional autoencoder for compressing CIFAR images