Abstract:Conventional detectors suffer from performance degradation when dealing with long-tailed data due to a classification bias towards the majority head categories. In this paper, we contend that the learning bias originates from two factors: 1) the unequal competition arising from the imbalanced distribution of foreground categories, and 2) the lack of sample diversity in tail categories. To tackle these issues, we introduce a unified framework called BAlanced CLassification (BACL), which enables adaptive rectification of inequalities caused by disparities in category distribution and dynamic intensification of sample diversities in a synchronized manner. Specifically, a novel foreground classification balance loss (FCBL) is developed to ameliorate the domination of head categories and shift attention to difficult-to-differentiate categories by introducing pairwise class-aware margins and auto-adjusted weight terms, respectively. This loss prevents the over-suppression of tail categories in the context of unequal competition. Moreover, we propose a dynamic feature hallucination module (FHM), which enhances the representation of tail categories in the feature space by synthesizing hallucinated samples to introduce additional data variances. In this divide-and-conquer approach, BACL sets a new state-of-the-art on the challenging LVIS benchmark with a decoupled training pipeline, surpassing vanilla Faster R-CNN with ResNet-50-FPN by 5.8% AP and 16.1% AP for overall and tail categories. Extensive experiments demonstrate that BACL consistently achieves performance improvements across various datasets with different backbones and architectures. Code and models are available at <a class="link-external link-https" href="https://github.com/Tianhao-Qi/BACL" rel="external noopener nofollow">this https URL</a>.

Towards Balanced Learning for Instance Recognition

Self-Balanced R-CNN for Instance Segmentation

Domain-Invariant Proposals Based on a Balanced Domain Classifier for Object Detection

DIM: Long-tailed Object Detection and Instance Segmentation via Dynamic Instance Memory

Bal-R$^2$CNN: High Quality Recurrent Object Detection with Balance Optimization

Balanced Classification: A Unified Framework for Long-Tailed Object Detection

Residual Objectness for Imbalance Reduction

Delving deep into the imbalance of positive proposals in two-stage object detection

Improving Real-world Object Detection Using Balanced Loss

Instance-based Dynamic Label Assignment for Object Detection

Not All Instances Contribute Equally: Instance-Adaptive Class Representation Learning for Few-Shot Visual Recognition

Decoupling Representation and Classifier for Long-Tailed Recognition

Realizing balanced object detection through prior location scale information and repulsive loss

A Robust Classifier for Unbalanced-Data Learning

Foreground-Background Imbalance Problem in Deep Object Detectors: A Review

Recognizing Partial Biometric Patterns.

Improving Training Instance Quality in Aerial Image Object Detection With a Sampling-Balance-Based Multistage Network

Improving imbalance classification via ensemble learning based on two-stage learning

Deep Structured Instance Graph for Distilling Object Detectors

Rebalancing Multi-Label Class-Incremental Learning

Instance Mining with Class Feature Banks for Weakly Supervised Object Detection.