RFAssigner: A Generic Label Assignment Strategy for Dense Object Detection

TL;DR

RFAssigner is a novel label assignment strategy that improves dense object detection by adaptively selecting positive samples across scales, leading to state-of-the-art results without extra modules.

Contribution

It introduces RFAssigner, a scale-aware assignment method using Gaussian Receptive Field distances to enhance multi-scale learning in dense detectors.

Findings

Achieves state-of-the-art performance across object scales.

Improves positive sample assignment for small objects.

Does not require auxiliary modules or heuristics.

Abstract

Label assignment is a critical component in training dense object detectors. State-of-the-art methods typically assign each training sample a positive and a negative weight, optimizing the assignment scheme during training. However, these strategies often assign an insufficient number of positive samples to small objects, leading to a scale imbalance during training. To address this limitation, we introduce RFAssigner, a novel assignment strategy designed to enhance the multi-scale learning capabilities of dense detectors. RFAssigner first establishes an initial set of positive samples using a point-based prior. It then leverages a Gaussian Receptive Field (GRF) distance to measure the similarity between the GRFs of unassigned candidate locations and the ground-truth objects. Based on this metric, RFAssigner adaptively selects supplementary positive samples from the unassigned pool, promoting a more balanced learning process across object scales. Comprehensive experiments on three datasets with distinct object scale distributions validate the effectiveness and generalizability of our method. Notably, a single FCOS-ResNet-50 detector equipped with RFAssigner achieves state-of-the-art performance across all object scales, consistently outperforming existing strategies without requiring auxiliary modules or heuristics.