Explore a thought-provoking conference talk that delves into the critical importance of documenting computer vision datasets and encourages the adoption of reflexive data practices. Presented at the FAccT 2021 virtual conference, this 18-minute research track presentation by M. Miceli, T. Yang, L. Naudts, M. Schuessler, D. Serbanescu, and A. Hanna challenges conventional approaches to dataset creation and management. Gain insights into the ethical considerations and potential biases inherent in computer vision datasets, and discover how proper documentation can lead to more transparent and accountable AI systems. Learn about innovative strategies for implementing reflexive data practices that can enhance the quality and reliability of machine learning models while addressing concerns related to fairness, accountability, and transparency in artificial intelligence.
Documenting Computer Vision Datasets - An Invitation to Reflexive Data Practices
Association for Computing Machinery (ACM) via YouTube
Overview
Syllabus
Documenting Computer Vision Datasets: An Invitation to Reflexive Data Practices
Taught by
ACM FAccT Conference
Related Courses
-
Towards Accountability for Machine Learning Datasets - Practices from Software Engineering and Infrastructure
-
Getting High-Quality Data for Your Computer Vision Models - Building Computer Vision Models
-
One Label, One Billion Faces - Usage and Consistency of Racial Categories in Computer Vision
-
Caring for Datasets - A Framework for Deprecating Datasets and Responsible Data Stewardship
-
Dataset Management for Computer Vision - Important Component to Delivering Computer Vision Solutions
-
Hiding Objects from Computer Vision by Exploiting Correlation Biases
