Event-Based Object Detection
Overview
Event-based object detection studies how to identify and localize objects from asynchronous brightness-change signals, which can capture fast motion with fine timing and little redundant data. Methods convert events into time surfaces, voxel or continuous-time representations, or process them directly with recurrent, sparse-attention, state-space, and spiking neural networks; RGB-event fusion is also used when appearance cues help. Automotive benchmarks such as Gen1 and 1Mpx support comparison, while pedestrian, drone, and small-object settings test performance under motion, low light, and sparse target evidence. A central challenge is balancing accuracy with latency and energy use while improving robustness across sensors and scenes, with benchmark coverage and real-world evaluation still important open issues.
Latest papers
36 papers1 paper in the last four weeks, down 75% on the four weeks before. 0.0% of all new papers.