← Back to the feed
Mobility & logisticsResearch findings

Generated rain and night scenes support a driving detector

Original published 2024.07.19 00:32 KSTAdded to Blackdeck 2026.10.18 15:12 KST

What has changed?

SUSTechGAN evaluates synthetic adverse-weather imagery as additional detector training data.

Key points

  1. The generator creates road scenes depicting rain or nighttime conditions.
  2. The study adds those images to training data and retrains a YOLOv5 detector.
  3. The authors report detection gains over the other generation approaches used in their comparison.

Connect it to your work

Driving teams can consider synthetic data for scarce conditions, followed by separate field evaluation.

What remains unproven

Detector improvements do not establish overall driving safety or performance on every unfamiliar road.

Originals used for this note

Links open the original publisher in a new tab.

  1. arXivSUSTechGAN: Image Generation for Object Detection in Adverse Conditions of Autonomous Driving