T-CLIP: Enabling Thermal Perception for Contrastive Language-Image Pretraining
Tayeba Qazi [email protected], Ayush Maheshwari [email protected], Prerana Mukherjee [email protected], Brejesh Lall [email protected] · 2026
Read T-CLIP: Enabling Thermal Perception for Contrastive Language-Image Pretraining by Tayeba Qazi [email protected], Ayush Maheshwari [email protected], Prerana Mukherjee [email protected], Brejesh Lall [email protected] (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-00673.txt.
Description
Qazi and co-authors teach computers to link heat images with words in this 2026 paper. Thermal cameras see heat, so they work in the dark and in bad weather. But CLIP, a well-known image-text model, fails on heat images. There are few captioned thermal datasets, and standard models grasp heat poorly.
The team builds IR-Cap, a captioning pipeline that uses physics to write both broad and fine notes on heat scenes. It covers three public test sets. Their model, T-CLIP, adapts CLIP with two separate add-on parts. One part learns whole-scene heat, and one learns the heat signs of single objects. The authors report gains over all baselines on three thermal tests of cross-modal search. They also show a first demo of text-guided thermal image creation.
Catalog record
| CREATOR | Tayeba Qazi [email protected], Ayush Maheshwari [email protected], Prerana Mukherjee [email protected], Brejesh Lall [email protected] |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 88 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.00673