Abstract: Clouds and their shadows hinder accurate analysis of optical remote sensing imagery, making cloud removal an indispensable preprocessing step in remote sensing. However, existing methods ...
TAEHV is a Tiny AutoEncoder for Hunyuan Video (and other similar video models). TAEHV can encode and decode latents into videos more cheaply (in time & memory) than the full-size video VAEs, at the ...
Abstract: Hyperspectral unmixing (HU) aims to decompose mixed pixels into their constituent spectral signatures and estimate their corresponding fractional abundances. Recently, the nonlinear spectral ...