A General Gaussian Heatmap Label Assignment for Arbitrary-Oriented Object Detection

TL;DR

This paper introduces a novel label assignment method using 2-D oriented Gaussian heatmaps for arbitrary-oriented object detection, improving shape and direction modeling with low tuning effort.

Contribution

It proposes a general Gaussian heatmap label assignment strategy that is anchor-free, adaptive, and applicable to various AOOD methods, enhancing detection accuracy.

Findings

Improves AOOD performance on public datasets

Reduces parameter-tuning and computational costs

Applicable to lightweight models on embedded platforms

Abstract

Recently, many arbitrary-oriented object detection (AOOD) methods have been proposed and attracted widespread attention in many fields. However, most of them are based on anchor-boxes or standard Gaussian heatmaps. Such label assignment strategy may not only fail to reflect the shape and direction characteristics of arbitrary-oriented objects, but also have high parameter-tuning efforts. In this paper, a novel AOOD method called General Gaussian Heatmap Label Assignment (GGHL) is proposed. Specifically, an anchor-free object-adaptation label assignment (OLA) strategy is presented to define the positive candidates based on two-dimensional (2-D) oriented Gaussian heatmaps, which reflect the shape and direction features of arbitrary-oriented objects. Based on OLA, an oriented-bounding-box (OBB) representation component (ORC) is developed to indicate OBBs and adjust the Gaussian center prior weights to fit the characteristics of different objects adaptively through neural network learning. Moreover, a joint-optimization loss (JOL) with area normalization and dynamic confidence weighting is designed to refine the misalign optimal results of different subtasks. Extensive experiments on public datasets demonstrate that the proposed GGHL improves the AOOD performance with low parameter-tuning and time costs. Furthermore, it is generally applicable to most AOOD methods to improve their performance including lightweight models on embedded platforms.