HashEncoding: Autoencoding with Multiscale Coordinate Hashing
This is an incremental improvement for autoencoder architectures in computer vision, potentially benefiting tasks like image reconstruction and optical flow.
The paper tackles the problem of reducing decoder parameters in autoencoders by introducing HashEncoding, which uses a multiscale coordinate hash function to enable a per-pixel decoder without convolutions, resulting in a smaller embedding space and improved generalizability.
We present HashEncoding, a novel autoencoding architecture that leverages a non-parametric multiscale coordinate hash function to facilitate a per-pixel decoder without convolutions. By leveraging the space-folding behaviour of hashing functions, HashEncoding allows for an inherently multiscale embedding space that remains much smaller than the original image. As a result, the decoder requires very few parameters compared with decoders in traditional autoencoders, approaching a non-parametric reconstruction of the original image and allowing for greater generalizability. Finally, by allowing backpropagation directly to the coordinate space, we show that HashEncoding can be exploited for geometric tasks such as optical flow.