Improved object detection method for autonomous driving based on DETR
Huaqi Zhao, Songnan Zhang, Xiang Peng, Zhengguang Lu, Guojing Li · 2025
Read Improved object detection method for autonomous driving based on DETR by Huaqi Zhao, Songnan Zhang, Xiang Peng, Zhengguang Lu, Guojing Li (2025) 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/pmc-11788285.txt.
Description
Huaqi Zhao and co-authors improve an object-spotting model for self-driving cars, in work published in 2024. Spotting people, cars, and signs fast and well is core to safe driving. Their base model uses a transformer design, which can struggle with objects at mixed scales.
Three upgrades were added. One pulls traits and place facts at several scales. One trims the focus step so it scans by rows and columns, which cuts compute and speeds output. One tunes loss weights on the fly to speed learning and lift accuracy. Tests on three public sets raised accuracy by about three to four points each, and frame rate rose by 84 percent.
Catalog record
| CREATOR | Huaqi Zhao, Songnan Zhang, Xiang Peng, Zhengguang Lu, Guojing Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |