Learning Spatial Fusion for Single-Shot Object Detection

TL;DR

This paper introduces adaptively spatial feature fusion (ASFF), a data-driven method to improve scale-invariance in single-shot object detection by effectively fusing features across scales with minimal additional computational cost.

Contribution

It proposes a novel ASFF strategy that learns to spatially filter conflicting information, enhancing feature consistency across scales in single-shot detectors like YOLOv3.

Findings

Achieves 38.1% AP at 60 FPS on MS COCO

Improves scale-invariance of features in object detection

Maintains nearly free inference overhead

Abstract

Pyramidal feature representation is the common practice to address the challenge of scale variation in object detection. However, the inconsistency across different feature scales is a primary limitation for the single-shot detectors based on feature pyramid. In this work, we propose a novel and data driven strategy for pyramidal feature fusion, referred to as adaptively spatial feature fusion (ASFF). It learns the way to spatially filter conflictive information to suppress the inconsistency, thus improving the scale-invariance of features, and introduces nearly free inference overhead. With the ASFF strategy and a solid baseline of YOLOv3, we achieve the best speed-accuracy trade-off on the MS COCO dataset, reporting 38.1% AP at 60 FPS, 42.4% AP at 45 FPS and 43.9% AP at 29 FPS. The code is available at https://github.com/ruinmessi/ASFF