240FPS Stereo Vision from Monocular Mixed Spikes
Monocular mixed spikes enable 240 FPS stereo vision by recovering disparity from temporally interleaved spike measurements.
Monocular mixed spikes enable 240 FPS stereo vision by recovering disparity from temporally interleaved spike measurements.
A rapidly rotating SpokeND filter produces multi-attenuated spikes that jointly support high frame rate and high dynamic range video reconstruction.
Differential focus volumes built from event streams prompt the network toward geometrically informative regions, improving dense metric depth accuracy and edge preservation.
A zero-shot diffusion framework reconstructs high-quality color video from chromatic spike streams and sub-millisecond intervals without paired training data.
Self-supervised denoising on real mosaicked chromatic spikes plus progressive warping for pseudo long-term exposure frames enables 2000 FPS color HDR reconstruction with reduced noise and motion blur.
A rolling-mixed-bit readout interleaves multi-bit spikings with single-bit streams, and RMB-Net fuses them via cross-bit and cross-time attention to reconstruct HDR and HFR video with less than 2% extra data volume.