ProFPN: Progressive feature pyramid network with soft proposal assignment for object detection

Junjie Ke, Lihuo He, Bo Han, Jie Li, Xinbo Gao

Abstract: Highlights•SPA leverages the scale of object proposals to generate RoI features.•Without adding computational costs, SPA boosts several detectors by 0.3 AP ∼<math><mo is="true">∼</mo></math> 2.1 AP.•ProFPN combines regional contextual details and semantic information in a progressive manner.•ProFPN boosts transformer-based detectors while reducing FLOPs.