Visual Bias

Visual bias in computer vision models refers to the tendency of these systems to unfairly favor certain visual features or demographics, leading to inaccurate or discriminatory outcomes. Current research focuses on identifying and mitigating these biases across various model architectures, including vision-language models and those used in tasks like audio-visual localization and zero-shot learning, often employing techniques like adversarial training and uncertainty-weighted loss functions. Understanding and addressing visual bias is crucial for ensuring fairness and reliability in AI systems, impacting applications ranging from facial recognition to medical image analysis and beyond. The development of robust bias detection and mitigation strategies is a key objective in advancing the field.

Papers

October 12, 2024

Debiasing Vison-Language Models with Text-Only Training
Yunfan Yang, Chaoquan Jiang, Zhiyu Lin, Jinlin Xiao, Jiaming Zhang, Jitao Sang
Large Language Model Vision Language Model Multi Target Visual Bias Text Only Training

August 25, 2024

Unveiling Visual Biases in Audio-Visual Localization Benchmarks
Liangyu Chen, Zihao Yue, Boshen Xu, Qin Jin
Audio Visual Learning Audio Visual Source Localization Visual Bias

June 25, 2024

Evaluating Fairness in Large Vision-Language Models Across Diverse Demographic Attributes and Prompts
Xuyang Wu, Yuan Wang, Hsin-Tai Wu, Zhiqiang Tao, Yi Fang
Large Vision Language Model Complex Prompt Visual Understanding Socio Demographic Fairness Analysis Visual Bias Fairness Benchmark Datasets

March 14, 2024

Are Vision Language Models Texture or Shape Biased and Can We Steer Them?
Paul Gavrikov, Jovita Lukasik, Steffen Jung, Robert Geirhos, Bianca Lamm, Muhammad Jehanzeb Mirza, Margret Keuper, Janis Keuper
Vision Language Model Vision Language Multimodal Model Shape Bias Visual Bias

January 15, 2024

Bias-Conflict Sample Synthesis and Adversarial Removal Debias Strategy for Temporal Sentence Grounding in Video
Zhaobo Qi, Yibo Yuan, Xiaowen Ruan, Shuhui Wang, Weigang Zhang, Qingming Huang
Source Video Temporal Sentence Grounding Adversarial Debiasing Bias Aligned Sample Visual Bias

December 26, 2023

Revealing the Proximate Long-Tail Distribution in Compositional Zero-Shot Learning
Chenyi Jiang, Haofeng Zhang
Class Imbalance Compositional Zero Shot Learning Long Tailed Distribution Object State Visual Bias

November 29, 2023

GELDA: A generative language annotation framework to reveal visual biases in datasets
Krish Kabra, Kathleen M. Lewis, Guha Balakrishnan
Data Set Vision Language Model Synthetic Data Generative Language Model Visual Attribute Visual Bias FAIR Datasets

May 31, 2023

Unveiling Cross Modality Bias in Visual Question Answering: A Causal View with Possible Worlds VQA
Ali Vosoughi, Shijian Deng, Songyang Zhang, Yapeng Tian, Chenliang Xu, Jiebo Luo
Visual Question Answering 3d Vqa Counterfactual Inference Language Bias VQA System Causal View Visual Bias

March 25, 2023

Fairness meets Cross-Domain Learning: a new perspective on Models and Metrics
Leonardo Iurada, Silvia Bucci, Timothy M. Hospedales, Tatiana Tommasi
Full Model Procedural Fairness Cross Domain Metric Library New Perspective Model Fairness Fairness Method Visual Bias

January 26, 2023

Discovering and Mitigating Visual Biases through Keyword Explanation
Younghyun Kim, Sangwoo Mo, Minkyu Kim, Kyungmin Lee, Jaeho Lee, Jinwoo Shin
Scientific Discovery Contextual Biasing Keyword Enrollment Bias Discovery Visual Bias

April 20, 2022

Epistemic Uncertainty-Weighted Loss for Visual Bias Mitigation
Rebecca S Stone, Nishant Ravikumar, Andrew J Bulpitt, David C Hogg
Adaptive Loss Potential Bias Bias Benchmark Visual Bias High Epistemic Uncertainty