FLAIR

FLAIR is a large-scale annotated image dataset containing 429,078 images from 51,414 Flickr users, designed to evaluate multi-label classification in the context of federated learning, capturing challenges like heterogeneous user data and long-tailed label distribution.

Papers using FLAIR (5)

Federated Learning for MRI-based BrainAGE: a multicenter study on post-stroke functional outcome prediction2025FLAIR: Federated Learning Annotated Image Repository2022 · 4 citesOn the Efficacy of Differentially Private Few-shot Image Classification2023 · 4 citesLanguage-Guided Transformer for Federated Multi-Label Classification2023HeteroSwitch: Characterizing and Taming System-Induced Data Heterogeneity in Federated Learning2024