A visible-light and thermal-infrared images datasets for dual-spectrum depth estimation
update: we will publish the dataset after our paper accepted
We use two hybrid RGB-Thermal cameras which can output optical-aligned visible-light(VIS) and thermal-infrared(TIR) images to build a dual-spectrum stereo camera. A Velodyne-64E LiDAR sensor is used for 3D point cloud to get ground-truth.
For our dual-spectrum stereo camera, each data frame contains six images: left/right VIS images, left/right TIR image, and left/right depth map ground-truth.
Datasets contains images with varing sences and illuminations.
