Thermal infrared face image data enhancement method based on deep learning

Ruihan Xu; Junnian Wang; Jisi Liu; Feng Ni; Bijia Cao

doi:10.1117/12.2673366

3 April 2023 Thermal infrared face image data enhancement method based on deep learning

Ruihan Xu, Junnian Wang, Jisi Liu, Feng Ni, Bijia Cao

Proceedings Volume 12599, Second International Conference on Digital Society and Intelligent Systems (DSInS 2022); 125990E (2023) https://doi.org/10.1117/12.2673366
Event: 2nd International Conference on Digital Society and Intelligent Systems (DSInS 2022), 2022, Chendgu, China

Abstract

The high accuracy of deep learning in various computer vision tasks often relies on massive image data, but there is usually not enough image data to complete the task in actual project applications. Aiming at the problem that thermal infrared face images are few and difficult to obtain in the absence of light, this paper proposes a method to generate thermal infrared face images based on color face images to obtain more thermal infrared face image data. This method takes the conditional generative adversarial (CGAN) network as the framework and uses the deep high-resolution network and the Patch Net network as the generator and discriminator to build the network model. By verifying the network model on the two datasets, it can be proved that the proposed method is clearer, and the details of the facial features are more obvious than the thermal infrared face images generated by the traditional image generation methods. Compared with the FID values of the original generation method, the FID values of the generated image using this method were reduced by 136 and 92. Therefore. The unsupervised data augmentation method based on HRNet network as a generator can also be applied to other visual tasks such as object recognition and detection that lack image data.

Citation Download Citation

Ruihan Xu, Junnian Wang, Jisi Liu, Feng Ni, and Bijia Cao "Thermal infrared face image data enhancement method based on deep learning", Proc. SPIE 12599, Second International Conference on Digital Society and Intelligent Systems (DSInS 2022), 125990E (3 April 2023); https://doi.org/10.1117/12.2673366

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available