Long-tailed Instance Segmentation using Gumbel Optimized Loss

Konstantinos Panagiotis Alexandridis, Jiankang Deng, Anh Nguyen, Shan Luo
7/22/2022
cs.CV

Official code: kostas1515/GOL (18 stars, Apache-2.0, last commit 2022-11-24).

Abstract

Major advancements have been made in the field of object detection and segmentation recently. However, when it comes to rare categories, the state-of-the-art methods fail to detect them, resulting in a significant performance gap between rare and frequent categories. In this paper, we identify that Sigmoid or Softmax functions used in deep detectors are a major reason for low performance and are sub-optimal for long-tailed detection and segmentation. To address this, we develop a Gumbel Optimized Loss (GOL), for long-tailed detection and segmentation. It aligns with the Gumbel distribution of rare classes in imbalanced datasets, considering the fact that most classes in long-tailed detection have low expected probability. The proposed GOL significantly outperforms the best state-of-the-art method by 1.1% on AP , and boosts the overall segmentation by 9.0% and detection by 8.0%, particularly improving detection of rare classes by 20.3%, compared to Mask-RCNN, on LVIS dataset. Code available at: https://github.com/kostas1515/GOL

View on arXivView PDF

Code Implementations(1)

kostas1515/GOLOfficial100%
Apache-2.0

[ECCV2022] Gumbel Optimised Loss for Long Tailed Instance Segmentation.

182Shell, PythonJul 14, 20223 years agoApache-2.0
gumbelinstance-segmentationlong-tailed-detectionlong-tailed-recognitionlvis+1 more

Similar Papers

Rectify the Regression Bias in Long-Tailed Object Detection

91% match2024

Improving Long-Tailed Object Detection with Balanced Group Softmax and Metric Learning

90% match2025

Fractal Calibration for long-tailed object detection

89% match1 repo62024

Learning from Rich Semantics and Coarse Locations for Long-tailed Object Detection

86% match1 repo82023

Boosting Long-tailed Object Detection via Step-wise Learning on Smooth-tail Data

85% match2023

PAT: Pixel-wise Adaptive Training for Long-tailed Segmentation

84% match2024

Cite this paper

@article{alexandridis2022longtailed,
  title  = {Long-tailed Instance Segmentation using Gumbel Optimized Loss},
  author = {Konstantinos Panagiotis Alexandridis and Jiankang Deng and Anh Nguyen and Shan Luo},
  year   = {2022},
  eprint = {2207.10936},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2207.10936v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2207.10936.svg)](https://codeofpaper.com/paper/2207.10936)

Discussion