Jianxiong Tang*

Spike-Temporal Latent Representation for Energy-Efficient Event-to-Video Reconstruction

"Event-to-Video (E2V) reconstruction aims to recover grayscale video from neuromorphic event streams, with Spiking Neural Networks (SNNs) being promising energy-efficient models for this task. Event voxels effectively compress event streams for E2V reconstruction, yet their temporal latent represent…