Point2RBox-v3: Self-Bootstrapping from Point Annotations via Integrated Pseudo-Label Refinement and Utilization

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Teng, Fan, Ziqian, Liu, Mingxin, Zhang, Xin, Lu, Xudong, Li, Wentong, Zhou, Yue, Yu, Yi, Li, Xiang, Yan, Junchi, Yang, Xue
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914297293045760
author Zhang, Teng
Fan, Ziqian
Liu, Mingxin
Zhang, Xin
Lu, Xudong
Li, Wentong
Zhou, Yue
Yu, Yi
Li, Xiang
Yan, Junchi
Yang, Xue
author_facet Zhang, Teng
Fan, Ziqian
Liu, Mingxin
Zhang, Xin
Lu, Xudong
Li, Wentong
Zhou, Yue
Yu, Yi
Li, Xiang
Yan, Junchi
Yang, Xue
contents Driven by the growing need for Oriented Object Detection (OOD), learning from point annotations under a weakly-supervised framework has emerged as a promising alternative to costly and laborious manual labeling. In this paper, we discuss two deficiencies in existing point-supervised methods: inefficient utilization and poor quality of pseudo labels. Therefore, we present Point2RBox-v3. At the core are two principles: 1) Progressive Label Assignment (PLA). It dynamically estimates instance sizes in a coarse yet intelligent manner at different stages of the training process, enabling the use of label assignment methods. 2) Prior-Guided Dynamic Mask Loss (PGDM-Loss). It is an enhancement of the Voronoi Watershed Loss from Point2RBox-v2, which overcomes the shortcomings of Watershed in its poor performance in sparse scenes and SAM's poor performance in dense scenes. To our knowledge, Point2RBox-v3 is the first model to employ dynamic pseudo labels for label assignment, and it creatively complements the advantages of SAM model with the watershed algorithm, which achieves excellent performance in both sparse and dense scenes. Our solution gives competitive performance, especially in scenarios with large variations in object size or sparse object occurrences: 66.09%/56.86%/41.28%/46.40%/19.60%/45.96% on DOTA-v1.0/DOTA-v1.5/DOTA-v2.0/DIOR/STAR/RSAR.
format Preprint
id arxiv_https___arxiv_org_abs_2509_26281
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Point2RBox-v3: Self-Bootstrapping from Point Annotations via Integrated Pseudo-Label Refinement and Utilization
Zhang, Teng
Fan, Ziqian
Liu, Mingxin
Zhang, Xin
Lu, Xudong
Li, Wentong
Zhou, Yue
Yu, Yi
Li, Xiang
Yan, Junchi
Yang, Xue
Computer Vision and Pattern Recognition
Artificial Intelligence
Driven by the growing need for Oriented Object Detection (OOD), learning from point annotations under a weakly-supervised framework has emerged as a promising alternative to costly and laborious manual labeling. In this paper, we discuss two deficiencies in existing point-supervised methods: inefficient utilization and poor quality of pseudo labels. Therefore, we present Point2RBox-v3. At the core are two principles: 1) Progressive Label Assignment (PLA). It dynamically estimates instance sizes in a coarse yet intelligent manner at different stages of the training process, enabling the use of label assignment methods. 2) Prior-Guided Dynamic Mask Loss (PGDM-Loss). It is an enhancement of the Voronoi Watershed Loss from Point2RBox-v2, which overcomes the shortcomings of Watershed in its poor performance in sparse scenes and SAM's poor performance in dense scenes. To our knowledge, Point2RBox-v3 is the first model to employ dynamic pseudo labels for label assignment, and it creatively complements the advantages of SAM model with the watershed algorithm, which achieves excellent performance in both sparse and dense scenes. Our solution gives competitive performance, especially in scenarios with large variations in object size or sparse object occurrences: 66.09%/56.86%/41.28%/46.40%/19.60%/45.96% on DOTA-v1.0/DOTA-v1.5/DOTA-v2.0/DIOR/STAR/RSAR.
title Point2RBox-v3: Self-Bootstrapping from Point Annotations via Integrated Pseudo-Label Refinement and Utilization
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2509.26281