GPSMamba

Official PyTorch implementation of the paper GPSMamba

Introduction

Infrared Image Super-Resolution (IRSR) is challenged by low-contrast, sparsely-textured data that demands robust global context modeling. While State-Space Models (SSMs) like Mamba offer efficient long-range dependency modeling, their intrinsic 1D causal scanning mechanism fragments the 2D image context, fundamentally limiting reconstruction fidelity. To address this, we present GPSMamba, a framework that systematically overcomes this limitation through a synergy of non-causal architectural prompting and global frequency-domain supervision. First, our Adaptive Semantic-Frequency State Space Module (ASF-SSM) embodies the non-causal prompting. It injects a dynamic prompt, derived from global frequency information, directly into the Mamba block. This breaks the rigid causal chain from within, enabling a non-causal, globally-aware state transition. Second, our Thermal-Spectral Attention and Phase Consistency (TSAPC) Loss provides explicit global supervision. Its thermal-spectral attention mechanism overcomes the limitations of patch-based scanning by selectively enforcing spectral fidelity on sparse thermal targets across non-adjacent patches. Through the synergy of these two innovations, extensive experimental results demonstrate that GPSMamba achieves state-of-the-art performance on benchmark IRSR datasets.

Approach overview

Requirements

• Python 3.9, PyTorch >= 1.11
• BasicSR 1.4.2
• Platforms: Ubuntu 18.04, cuda-11

Installation

Clone the repo

git clone https://github.com/yongsongH/GPSMamba.git

Install dependent packages

cd GPSMamba

pip install -r requirements.txt

Install BasicSR

python setup.py develop

You can also refer to this INSTALL.md for installation

Dataset prepare

Please check this page.

Model

Pre-trained models can be downloaded from this link.

Evaluation

Training

• Run the following commands for training:

python basicsr/train.py -opt options/train/train_GPSMamba_IRSR_x2.yml

python basicsr/train.py -opt options/train/train_GPSMamba_IRSR_x4.yml

Testing

---

Run

  python basicsr/test.py -opt options/test/GPSMamba/test_GPSMamba_x4.yml

  python basicsr/test.py -opt options/test/GPSMamba/test_GPSMamba_x2.yml

Setup

For environment setup, you can also refer to these projects:

• BasicSR-based: IRSRMamba
• Mamba-based: MambaIR

Thanks for their awesome work.

Contact

If you meet problems, please describe them and contact me.

Impolite or anonymous emails are not welcome. There may be some difficulties for me to respond to the email without self-introduce. Thank you for understanding.

Acknowledgement

This work is under peer review. The updated manuscript and dataset will be released after the paper is accepted.

---

Legal Action and Patent Clause

This project is licensed under the Apache License, Version 2.0. A copy of the Apache License, Version 2.0, is included in the LICENSE file in this repository. You may also obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0.

Any violation of the terms of the Apache License, Version 2.0, may result in legal action and liability for damages.