Projects Computer Vision Gait Recognition via Invertible Koopman AutoencoderWe model gait in a Koopman embedding space to obtain interpretable motion descriptors and employ a lightweight reversible autoencoder.
