Learning to recover weak signals: A two-stage graph-based framework for all-weather RGB-T object detection
Ruoyan Pei, Pengge Ma, Jinwang Qian, Hai Yang, Junling Sun · 2026
Read Learning to recover weak signals: A two-stage graph-based framework for all-weather RGB-T object detection by Ruoyan Pei, Pengge Ma, Jinwang Qian, Hai Yang, Junling Sun (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/plos-pone-0354466.txt.
Description
Ruoyan Pei and colleagues present in 2026 a two stage detection system for finding road users in poor light, blending normal camera and heat images. Distant small objects give weak signals, and glare or darkness degrades each image type in its own way. The first stage fuses the two views with gated attention, and the second links candidate objects in a graph to reason about their layout.
Tests on two public datasets showed much stronger results across day and night, with the day to night gap shrinking dramatically. Small object detection improved a lot, and speed stayed fast enough for real time use. Self driving systems in hard conditions could benefit.
Catalog record
| CREATOR | Ruoyan Pei, Pengge Ma, Jinwang Qian, Hai Yang, Junling Sun |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80201 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |