CUHK-PEDES

The 'CUHK-PEDES' dataset/benchmark contains a collection of person images paired with natural language descriptions, and it is used to evaluate the effectiveness of methods in Text-Based Person Search (TBPS).

Papers using CUHK-PEDES (4)

Harnessing Weak Pair Uncertainty for Text-based Person Search2026DiCo: Disentangled Concept Representation for Text-to-image Person Re-identification2026Unified Multi-Dataset Training for TBPS2026SCMM: Calibrating Cross-modal Representations for Text-Based Person Search2023 · 6 cites