CSA: Cross-scale alignment with adaptive semantic aggregation and filter for image-text retrieval

Open Webpage

Zheng Liu, Junhao Xu, Shanshan Gao, Zhumin Chen

Published: 2025, Last Modified: 14 Jan 2026Pattern Recognit. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Proposes a cross-scale alignment framework without scale constraints.•Generates scale-adaptable semantic units by adaptive semantic aggregation.•Ensures semantic consistency with Position- and Co-occurrence-aware subsequences.•Filters out weak semantic associations by adaptive semantic filter.•Learns accurate image-text similarity by semantic unit alignment.