ReSAM: Refine, Requery, and Reinforce: Self-Prompting Point-Supervised Segmentation for Remote Sensing Images

arXiv — cs.CVFriday, December 12, 2025 at 5:00:00 AM
  • A new self-prompting, point-supervised framework has been proposed to enhance the Segment Anything Model (SAM) for remote sensing imagery, addressing challenges related to domain shifts and sparse annotations. This method utilizes a Refine-Requery-Reinforce loop to improve segmentation quality without requiring full-mask supervision, demonstrating effectiveness on benchmark datasets like WHU, HRSID, and NWPU VHR-10.
  • This development is significant as it enhances SAM's capabilities in remote sensing applications, which have previously suffered from limited performance due to the lack of dense annotations. By leveraging sparse point annotations, the new framework aims to make SAM more robust and effective in diverse imaging scenarios.
  • The advancement reflects a broader trend in AI and image segmentation, where models are increasingly being adapted for specific tasks through innovative techniques like self-supervised learning and parameter-efficient fine-tuning. This evolution highlights the ongoing efforts to improve model efficiency and adaptability across various domains, including medical imaging and agricultural applications.
— via World Pulse Now AI Editorial System

Was this article worth reading? Share it

Recommended apps based on your readingExplore all apps
Continue Readings
Sesame Plant Segmentation Dataset: A YOLO Formatted Annotated Dataset
PositiveArtificial Intelligence
A new dataset, the Sesame Plant Segmentation Dataset, has been introduced, featuring 206 training images, 43 validation images, and 43 test images formatted for YOLO segmentation. This dataset focuses on sesame plants at early growth stages, captured under various environmental conditions in Nigeria, and annotated with the Segment Anything Model version 2.

Ready to build your own newsroom?

Subscribe to unlock a personalised feed, podcasts, newsletters, and notifications tailored to the topics you actually care about