Authors: Peiqi Xie

Description: A novel Half-To-All Multiple Object Tracking (HTAMotr) approach is proposed to address the challenges posed by incomplete annotation in video text tracking. Three key strategies were introduced:} rotated queries to improve anchor alignment with text regions, the Proposal-For-Groundtruth Strong Correlation (PForG) strategy to mitigate the negative effects of incomplete annotations, and an overlapping anchor filter to resolve ID switching issues. Experiments conducted on the DSText dataset demonstrate the effectiveness of HTAMotr, achieving state-of-the-art performance without requiring additional pre-training data or extensive epochs. By addressing the limitations of traditional MOTR paradigms, this work contributes to advancing video text tracking techniques and facilitating the development of more robust and efficient algorithms.

Authors: Hongen Liu, Di Sun, Jiahao Wang, Yi Liu, Gang Pan

Affiliation: College of Intelligence and Computing, Tianjin University；Tianjin University of Science and Technology; Baidu Inc.

Description: We propose a Language Collaboration and Glyph Perception Model, termed LOGO to enhance the performance of conventional text spotters through the integration of a synergy module. To achieve this goal, a language synergy classifier (LSC) is designed to explicitly discern text instances from background noise in the recognition stage. Besides, the glyph supervision and visual position mixture module are proposed to enhance the recognition accuracy of noisy text regions, and acquire more discriminative tracking features, respectively.