Lightweight Backbone Networks Only Require Adaptive Lightweight Self-Attention Mechanisms

Benchmark Model Rank Results
real-time-object-detection-on-coco-val2017LOLViT-S (YOLO head)–box AP: 37.8FPS (batch 1, GPU in brackets): 113 (RTX 4060 Ti)
real-time-object-detection-on-coco-val2017LOLViT-X (YOLO head)–box AP: 41.9FPS (batch 1, GPU in brackets): 99 (RTX 4060 Ti)