You Only Look Omni Gradient Backpropagation for Moving Infrared Small Target Detection

2025-11-17

AI Summary

Mobile infrared small target detection is highly challenging due to low signal-to-clutter ratio, extreme target-background imbalance, and weak feature discriminability. Existing approaches primarily focus on spatiotemporal modeling, yet their performance bottlenecks stem fundamentally from ambiguous single-frame feature representation. To address this, we propose a backpropagation-driven feature pyramid network (BP-FPN), the first framework to reformulate feature learning from a gradient-flow perspective. Specifically, we introduce Gradient-Isolated Low-level Shortcuts (GILS) to mitigate interference from shortcut connections, and Directional Gradient Regularization (DGR) to enhance response consistency for faint targets. This design significantly improves both the efficiency and robustness of fine-grained feature fusion. Extensive experiments on multiple public benchmarks demonstrate state-of-the-art performance. The model is lightweight and plug-and-play, validating its effectiveness and generalizability across diverse scenarios.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Computer Vision: Low Level & Physics-based Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

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

Abstract

Moving infrared small target detection is a key component of infrared search and tracking systems, yet it remains extremely challenging due to low signal-to-clutter ratios, severe target-background imbalance, and weak discriminative features. Existing deep learning methods primarily focus on spatio-temporal feature aggregation, but their gains are limited, revealing that the fundamental bottleneck lies in ambiguous per-frame feature representations rather than spatio-temporal modeling itself. Motivated by this insight, we propose BP-FPN, a backpropagation-driven feature pyramid architecture that fundamentally rethinks feature learning for small target. BP-FPN introduces Gradient-Isolated Low-Level Shortcut (GILS) to efficiently incorporate fine-grained target details without inducing shortcut learning, and Directional Gradient Regularization (DGR) to enforce hierarchical feature consistency during backpropagation. The design is theoretically grounded, introduces negligible computational overhead, and can be seamlessly integrated into existing frameworks. Extensive experiments on multiple public datasets show that BP-FPN consistently establishes new state-of-the-art performance. To the best of our knowledge, it is the first FPN designed for this task entirely from the backpropagation perspective.

Problem

• Detecting moving infrared small targets with low signal-to-clutter ratios
• Addressing severe target-background imbalance in infrared detection
• Improving ambiguous per-frame feature representations for small targets

Innovation

• Gradient-Isolated Low-Level Shortcut for fine-grained details
• Directional Gradient Regularization for hierarchical consistency
• Backpropagation-driven feature pyramid architecture for targets