Energy & materialsResearch findings
A Method for Addressing Data Shift in Autonomous Driving Object Detection
Original published 2025.08.16 10:52 KSTAdded to Blackdeck 2026.11.08 03:26 KST
What has changed?
Systematically analyzes data shift phenomena and proposes a model that accounts for dynamic distribution changes over time
Key points
- Analyzes how dynamic data distribution shifts affect object detection performance in autonomous driving systems
- Proposes a model that systematically accounts for time-varying data distribution changes
- Achieves superior performance on BDD100K dataset by integrating CycleGAN-based data augmentation into YOLOv5
Connect it to your work
Integrates CycleGAN-based data augmentation into the YOLOv5 framework to improve robustness against distribution shifts
What remains unproven
This research note is based on an abstract; check the original for test conditions and application scope.
Originals used for this note
Links open the original publisher in a new tab.