Pursuing Better Decision Boundaries for Long-Tailed Object Detection via Category Information Amount

TL;DR

This paper introduces the concept of category information amount to better understand category bias in object detection, proposing a new loss function that dynamically adjusts decision boundaries based on this measure, improving performance on long-tail and balanced datasets.

Contribution

It defines and measures category information amount, revealing its correlation with detection difficulty, and proposes IGAM Loss to adapt decision boundaries accordingly, reducing category bias.

Findings

IGAM Loss improves detection on long-tail datasets like LVIS and COCO-LT.

Category information amount correlates negatively with detection accuracy.

Method enhances underrepresented categories in balanced datasets like Pascal VOC.

Abstract

In object detection, the instance count is typically used to define whether a dataset exhibits a long-tail distribution, implicitly assuming that models will underperform on categories with fewer instances. This assumption has led to extensive research on category bias in datasets with imbalanced instance counts. However, models still exhibit category bias even in datasets where instance counts are relatively balanced, clearly indicating that instance count alone cannot explain this phenomenon. In this work, we first introduce the concept and measurement of category information amount. We observe a significant negative correlation between category information amount and accuracy, suggesting that category information amount more accurately reflects the learning difficulty of a category. Based on this observation, we propose Information Amount-Guided Angular Margin (IGAM) Loss. The core idea of IGAM is to dynamically adjust the decision space of each category based on its information amount, thereby reducing category bias in long-tail datasets. IGAM Loss not only performs well on long-tailed benchmark datasets such as LVIS v1.0 and COCO-LT but also shows significant improvement for underrepresented categories in the non-long-tailed dataset Pascal VOC. Comprehensive experiments demonstrate the potential of category information amount as a tool and the generality of our proposed method.