FLIR-IISR Dataset

This repository contains the FLIR-IISR dataset, a real-world infrared image super-resolution benchmark introduced in the paper Toward Real-world Infrared Image Super-Resolution: A Unified Autoregressive Framework and Benchmark Dataset.

Paper | Code

Dataset Details

FLIR-IISR is a real-world Infrared Image Super-Resolution (IISR) dataset consisting of 1,457 paired low-resolution (LR) and high-resolution (HR) infrared images. The data was acquired via automated focus variation and motion-induced blur to capture coupled optical and sensing degradations found in real-world conditions.

Composition

• Total number of image pairs: 1,457
• Image size: 1024 × 768
• Degradation labels:
  • Optical blur (1,305 pairs)
  • Motion blur (152 pairs)

Scene Categories (12 categories)

The dataset covers a wide variety of scenes:

• Building: 706
• Complex scene: 401
• Person: 309
• Regular object: 248
• Car: 234
• Statue: 157
• Road: 132
• Plane: 54
• Motorcycle: 27
• Bicycle: 22
• Tricycle: 13
• Bus: 5

Citation

@inproceedings{zou2026toward,
  title={Toward real-world infrared image super-resolution: A unified autoregressive framework and benchmark dataset},
  author={Zou, Yang and Ma, Jun and Jiao, Zhidong and Li, Xingyuan and Jiang, Zhiying and Liu, Jinyuan},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={16365--16375},
  year={2026}
}