Merging event cameras and image sensors to brighten dark photos

Low-light photography is hard: sensors miss detail in shadows, and event cameras (which capture motion extremely fast) are noisy in the dark. This work fuses both sensor types, using event cameras to recover high-contrast detail while filtering out their noise based on image brightness. The team built a custom dual-camera rig to collect 1,000+ synchronized event-image pairs from real scenes, then trained a model that outperforms existing methods across five datasets. Code and data are publicly available.