{
  "id": "2308.02213",
  "version": "v1",
  "published": "2023-08-04T09:11:07.000Z",
  "updated": "2023-08-04T09:11:07.000Z",
  "title": "Balanced Classification: A Unified Framework for Long-Tailed Object Detection",
  "authors": [
    "Tianhao Qi",
    "Hongtao Xie",
    "Pandeng Li",
    "Jiannan Ge",
    "Yongdong Zhang"
  ],
  "comment": "Accepted by IEEE Transactions on Multimedia, to be published; Code: https://github.com/Tianhao-Qi/BACL",
  "categories": [
    "cs.CV"
  ],
  "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 https://github.com/Tianhao-Qi/BACL.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2023-08-04T09:11:07.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "long-tailed object detection",
      "unified framework",
      "balanced classification",
      "tail categories",
      "novel foreground classification balance loss"
    ],
    "tags": [
      "github project"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}