News Release

Breaking new ground in aerial imaging: the AVIID dataset and visible-to-infrared image translation

Peer-Reviewed Publication

Aerospace Information Research Institute, Chinese Academy of Sciences

Overview of image-to-image translation methods that could be applied to aerial visible-to-infrared image translation. Each color represents a category.


Overview of image-to-image translation methods that could be applied to aerial visible-to-infrared image translation. Each color represents a category.

view more 

Credit: Journal of Remote Sensing

The rapid development of infrared technology, especially infrared cameras on UAVs, has expanded the applications of aerial infrared photography in military, industrial, agricultural, and environmental contexts. However, acquiring specific aerial infrared images is challenging due to high costs and photography limitations. Traditionally, infrared simulation software is used, which, despite being cost-effective and adjustable, has some notable weakness including low simulation accuracy and complex processing.

A study published on 10 November 2023, in the journal of Journal of Remote Sensing, has made significant strides in aerial visible-to-infrared image translation. It offers advantages like lower cost, higher efficiency, and enhanced downstream task performance, addressing issues like lack of datasets, methodological surveys, and comprehensive evaluation systems for image quality.

The research involved an in-depth analysis of image translation methodologies applied to the AVIID dataset. The team focused on the conversion of visible images to infrared images using various image-to-image translation techniques. They evaluated these techniques based on their effectiveness in generating high-quality infrared images. The study not only benchmarked existing methods but also analyzed key technologies for performance improvement. This included assessing the fidelity of the translated images in replicating true infrared imagery, as well as their utility in practical applications such as environmental monitoring and surveillance. The detailed analysis provided a solid foundation for future advancements in aerial image translation technology.

Professor Shaohui Mei, the lead researcher, stated, "Our dataset and evaluation system mark a significant step in aerial image translation, offering researchers a unique resource to develop and evaluate advanced algorithms in this field."

The team plans to extend their dataset and refine their evaluation system. They aim to address challenges like improving image quality under diverse conditions and integrating these technologies into real-world applications.

This research has potential applications in surveillance, environmental monitoring, and disaster response, where rapid and accurate image translation is critical. The dataset and methodologies developed could significantly reduce costs and enhance the capabilities of infrared imaging technologies.





Original Source URL

Funding information

The National Natural Science Foundation of China (62271409, 62171381).

About Journal of Remote Sensing

The Journal of Remote Sensing, an online-only Open Access journal published in association with AIR-CAS, promotes the theory, science, and technology of remote sensing, as well as interdisciplinary research within earth and information science.

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.