FS-DVS: A Frequency-Selective Dynamic Visual Sensing Paradigm for Enhancing Information Completeness
Researchers have developed FS-DVS, a new paradigm for dynamic vision sensors that mimics biological retinal ganglion cells. This approach integrates a learnable spatial filter before event triggering to improve contrast sensitivity and capture mid-spatial frequencies, addressing information loss in conventional sensors. The system demonstrates enhanced performance in object detection and action recognition, offering a noise-resilient and biologically plausible design for future neuromorphic sensors. AI
IMPACT Introduces a biologically inspired approach to sensor data processing, potentially improving downstream AI task performance.