An Approach to Synthesize Thermal Infrared Ship Images

Doan Thinh Vo, Phan Anh Đức, Nguyen Nhu Thao, Huong Ninh

Keywords: Thermal Infrared, Synthetic Data, Image to Image Translation

TL;DR: In this paper, we have presented a new approach based on I2I translation and segmentation model, which can directly and accurately create synthetic thermal infrared images from visible images.

Abstract: In this paper, we propose a method to produce synthetic thermal infrared (TIR) images using a diffusion-based image-to-image translation model. The model translates the abundantly available RGB images into synthetic TIR data closer to the domain of authentic TIR images. For this purpose, we explore the usage of an unpaired image translation neural model based on Schrödinger bridge algorithms. In addition, the visual characteristic of the object in the image is an important consideration when generating the results. Thus, we take advantage of a segmentation module before the image-to-image translation model to discriminate the background and object regions. We practice the model's performance with a self-proposed dataset comprising unpaired realistic RGB-TIR images. When incorporated into the training set, the synthesized images of our model significantly increase the classification accuracy by 15% and F1-score by 18% when only using realistic TIR images.

Submission Number: 51