RPCASSM: Robust PCA State Space Model For Infrared Small Target Detection

2026-06-01 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Computer Vision · Depth: Expert, quick

Summary

The RPCASSM network, published on 2026-06-01, addresses the challenge of infrared small target detection and segmentation, crucial for surveillance and maritime rescue. Mainstream visual state space models are inefficient and struggle to accurately model target edges due to low occupancy. RPCASSM proposes a novel architecture based on the Robust Principal Component Analysis (RPCA) paradigm, specifically designed for infrared small target properties in the spatial domain. It incorporates a Background State Space Module (BSSM) that uses a spatial probe scanning mechanism (SPCM) to model background information from spatial heterogeneous signals. Concurrently, a Target State Space Module (TSSM) employs a deformable prompt scanning mechanism (DPCM) to focus on the target's deformable space, leveraging its sparsity and local highlight. This design effectively overcomes limitations of existing models in accurately capturing infrared small target edge structures, with experimental results on benchmark datasets confirming its effectiveness. The code will be publicly available.

Key takeaway

For Computer Vision Engineers developing robust infrared small target detection systems, RPCASSM provides a specialized solution to overcome the limitations of mainstream visual state space models. You should consider integrating its Robust PCA-based Background State Space Module and Target State Space Module to accurately model target edges in low-occupancy scenarios. This approach promises enhanced performance for surveillance, security, and maritime rescue applications, with public code availability facilitating practical evaluation.

Key insights

RPCASSM leverages Robust PCA to design specialized state space modules for background and target, accurately detecting infrared small targets by their spatial properties.

Principles

• Infrared small targets require specialized spatial domain modeling.
• Decomposing background and target modeling improves detection accuracy.
• Target sparsity and local highlight are key properties for modeling.

Method

RPCASSM designs a Background State Space Module (BSSM) with a spatial probe scanning mechanism (SPCM) and a Target State Space Module (TSSM) with a deformable prompt scanning mechanism (DPCM).

In practice

• Apply to surveillance and security systems.
• Enhance maritime rescue operations.
• Improve infrared small target segmentation.

Topics

• Infrared Small Target Detection
• Robust Principal Component Analysis
• State Space Models
• Computer Vision
• Target Segmentation
• Surveillance

Code references

• PepperCS/RPCASSM

Best for: Research Scientist, AI Scientist, Computer Vision Engineer

Related on AIssential

• Object detection with Amazon Nova 2 Lite — Amazon Nova 2 Lite enables zero-shot object detection via natural language prompts, simplifying computer vision deployment.
• Honey, I Shrunk the Arc de Triomphe! — A new dataset and method mitigate scale-collapse in monocular depth estimation for distant, unconstrained scenes.
• TROPHIES: Temporal Reconstruction of Places, Humans, and Cameras from Multi-view Videos — TROPHIES unifies human, scene, and camera reconstruction from multi-view videos into a globally consistent 4D space.
• Edge-aware Decoding for Neural Asymmetric Routing — Neural asymmetric routing models improve by explicitly exposing transition-level edge information in the decoder, resolving representation-decision mismatch.
• LALE: Lightweight-Transformer Architecture for Land-Cover Estimation — LALE efficiently segments remote sensing imagery by combining ConvMixers for local detail and transformers for global context in a bifurcated encoder.
• Beyond Low-Rank: Low-Rank Sparse Prompting via Spiking Neural Network and Prompt Factorization — LoRSP uses spiking neural networks and low-rank factorization to create sparse, efficient visual prompts for vision models.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.