Deep learning based infrared small object segmentation: Challenges and future directions

TL;DR

This survey analyzes the challenges of infrared long-distance perception for small-object segmentation and surveys a wide range of deep-learning approaches. It frames the problem in terms of feature modeling, training data limitations, and cross-domain transfer, and reviews methods that advance local-detail preservation, multi-scale context, and semantic refinement within encoder-decoder architectures. It highlights the state-of-the-art performance on several infrared datasets while noting the trade-offs between accuracy and efficiency, and it emphasizes the need for infrared-focused pre-training and foundation-model-based transfer to achieve robust generalization. The paper offers future directions toward infrared-centric priors, efficient context modeling, and scalable pre-training to enable practical deployment in diverse long-distance scenarios.

Abstract

Infrared sensing is a core method for supporting unmanned systems, such as autonomous vehicles and drones. Recently, infrared sensors have been widely deployed on mobile and stationary platforms for detection and classification of objects from long distances and in wide field of views. Given its success in the vision image analysis domain, deep learning has also been applied for object recognition in infrared images. However, techniques that have proven successful in visible light perception face new challenges in the infrared domain. These challenges include extremely low signal-to-noise ratios in infrared images, very small and blurred objects of interest, and limited availability of labeled/unlabeled training data due to the specialized nature of infrared sensors. Numerous methods have been proposed in the literature for the detection and classification of small objects in infrared images achieving varied levels of success. There is a need for a survey paper that critically analyzes existing techniques in this domain, identifies unsolved challenges and provides future research directions. This paper fills the gap and offers a concise and insightful review of deep learning-based methods. It also identifies the challenges faced by existing infrared object segmentation methods and provides a structured review of existing infrared perception methods from the perspective of these challenges and highlights the motivations behind the various approaches. Finally, this review suggests promising future directions based on recent advancements within this domain.

Figures (12)

• Figure 1: The great challenges towards the goal of achieving high-quality infrared small object recognition of diverse long-distance perception.
• Figure 2: The architecture of this review paper constructed from the perspective of challenges.
• Figure 3: The commonly used encoder-decoder structure as well as the refinement strategies for infrared small object recognition.
• Figure 4: Commonly used attention models in infrared perception.
• Figure 5: Commonly used feature pyramids in deep learning.