Robustness analysis of YOLO and faster R-CNN for object detection in realistic weather scenarios with noise augmentation
Sana Gholinavaz, Nima saeedi, Sina Samadi Gharehveran · 2025
Read Robustness analysis of YOLO and faster R-CNN for object detection in realistic weather scenarios with noise augmentation by Sana Gholinavaz, Nima saeedi, Sina Samadi Gharehveran (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-12749240.txt.
Description
Sana Gholinavaz, Nima Saeedi, and Sina Samadi Gharehveran stress-test four object detectors against bad weather and digital noise, in a study published in 2025. Self-driving and traffic systems depend on spotting vehicles and pedestrians reliably, yet fog, rain, snow, and sandstorms degrade camera images badly. The team used a thousand-image benchmark of real adverse weather and layered on artificial fog, blur, and two kinds of noise.
On clean images YOLOv8m was most accurate. Under noise, the older Faster R-CNN design proved the most resilient, while the tiny YOLOv10n balanced speed and robustness well. Scores used standard precision and overlap measures throughout. The authors argue training pipelines should adapt to conditions, and benchmarks should reflect the weather systems will actually face.
Catalog record
| CREATOR | Sana Gholinavaz, Nima saeedi, Sina Samadi Gharehveran |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |