Deep Thermal Imaging Dataset
Introduced by Cho et al. in Deep Thermal Imaging: Proximate Material Type Recognition in the Wild through Deep Learning of Spatial Surface Temperature Patterns
The Deep Thermal Imaging dataset consists of two main datasets:
-
DeepTherm I (Indoor materials) - 15 indoor materials were used to create the dataset DeepTherm I which consists of 14,860 processed thermal images (average count of data for each individual class: 990.7, SD=425.9; 400-600 images of each material per each variable). The dataset was created by recording thermal image sequences in a room with different lighting levels (bright / dark), with/without air-conditioning, different places (on a floor or a desk) and from different perspectives (Figure 5). The spatial temperature patterns were collected from different angles and different distances (between 10 and 50 cm, from the camera lens to the material). The data was collected five times in about 3 weeks.
-
DeepTherm II (Outdoor materials) - 17 outdoor materials were targeted. The data collection process produced the DeepTherm II dataset which includes 26,584 labelled thermal images. The average number of collected spatial thermal patterns from each material was 1563.8 (SD=295.3; about 300-500 images of each material per each condition).
Papers
| Paper | Code | Results | Date | Stars |
|---|
Dataset Loaders
No data loaders found. You can
submit your data loader here.
Tasks
