Point-to-Mask: From Arbitrary Point Annotations to Mask-Level Infrared Small Target Detection

Dimensions overall score 8.0

• The proposed Point-to-Mask framework bridges low-cost point supervision and mask-level detection for infrared small target detection.method 95%
• The Physics-driven Adaptive Mask Generation (PAMG) module converts point annotations into compact target masks and geometric cues.method 95%
• The Radius-aware Point Regression Network (RPR-Net) reformulates IRSTD as target center localization and effective radius regression using spatiotemporal motion cues.method 95%
• The framework achieves strong pseudo-label quality, high detection accuracy, and efficient inference, approaching full-supervision performance under point-supervised settings.result 90%
• The method substantially reduces annotation cost compared to pixel-level segmentation approaches.technical 90%
• The approach constructs SIRSTD-Pixel, a sequential dataset with refined pixel-level annotations for systematic evaluation.method 95%
• The framework forms a closed loop where PAMG generates pseudo masks during training and RPR-Net predictions feed back to PAMG for pixel-level mask recovery during inference.method 90%
• Pixel-level segmentation approaches require costly dense annotations and are not well suited to tiny targets with weak texture and ambiguous boundaries.limitation 90%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card