Hybrid Mask Generation for Infrared Small Target Detection with Single-Point Supervision

2024-09-06

arXiv.org

AI Summary

To address the high annotation cost and low-quality mask generation under single-point supervision in infrared small target detection (SIRST), this paper proposes the first two-stage point-to-mask weakly supervised paradigm (point → box → mask). Methodologically: (1) a geometry-rule-driven Points-to-Box module robustly maps sparse point annotations to candidate bounding boxes; (2) a lightweight CNN architecture performs fine-grained Box-to-Mask prediction; (3) a self-training mechanism iteratively refines pseudo-masks by integrating handcrafted priors with data-driven features. Evaluated on three infrared datasets, the method achieves state-of-the-art performance in both mask IoU and detection mAP, significantly outperforming existing point-supervised approaches. It is the first to enable high-quality, robust pixel-level localization of infrared small targets using only single-point annotations.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Single-frame infrared small target (SIRST) detection poses a significant challenge due to the requirement to discern minute targets amidst complex infrared background clutter. Recently, deep learning approaches have shown promising results in this domain. However, these methods heavily rely on extensive manual annotations, which are particularly cumbersome and resource-intensive for infrared small targets owing to their minute sizes. To address this limitation, we introduce a Hybrid Mask Generation (HMG) approach that recovers high-quality masks for each target from only a single-point label for network training. Specifically, our HMG approach consists of a handcrafted Points-to-Mask Generation strategy coupled with a pseudo mask updating strategy to recover and refine pseudo masks from point labels. The Points-to-Mask Generation strategy divides two distinct stages: Points-to-Box conversion, where individual point labels are transformed into bounding boxes, and subsequently, Box-to-Mask prediction, where these bounding boxes are elaborated into precise masks. The mask updating strategy integrates the complementary strengths of handcrafted and deep-learning algorithms to iteratively refine the initial pseudo masks. Experimental results across three datasets demonstrate that our method outperforms the existing methods for infrared small target detection with single-point supervision.

Problem

• Detects infrared small targets in complex backgrounds
• Uses single-point supervision for high-quality pseudo masks
• Combines learning-free and learning-based methods to improve accuracy

Innovation

• Hybrid learning-free and learning-based methods
• Generates pseudo masks from point-level annotations
• Filters false alarms and retrieves missed detections