事件表示 | ViLab https://vilab.team/tag/%E4%BA%8B%E4%BB%B6%E8%A1%A8%E7%A4%BA/ 事件表示 Hugo Blox Builder (https://hugoblox.com)en-usMon, 26 Jun 2023 00:00:00 +0000 https://vilab.team/media/icon_hu2896232876136423579.png 事件表示 https://vilab.team/tag/%E4%BA%8B%E4%BB%B6%E8%A1%A8%E7%A4%BA/ Better and faster: Adaptive event conversion for event-based object detection https://vilab.team/publication/better-and-faster-adaptive-event-conversion-for-event-based-/ Mon, 26 Jun 2023 00:00:00 +0000 https://vilab.team/publication/better-and-faster-adaptive-event-conversion-for-event-based-/ <p>本文针对事件相机目标检测任务，提出高效事件表示Hyper Histogram，充分保留事件极性与时间信息；设计自适应事件转换模块AEC，基于事件密度通过自适应队列将事件流转换为超直方图，并适配现有帧基检测器；还提出事件增强方法Shadow Mosaic，提升样本多样性与泛化能力。在YOLOv5、Deformable-DETR和RetinaNet上验证，在1Mpx、Gen1和MVSEC-NIGHTL21数据集上取得显著优势，且推理速度快。</p>