Technical Report for ICRA 2025 GOOSE 2D Semantic Segmentation Challenge: Boosting Off-Road Segmentation via Photometric Distortion and Exponential Moving Average
This work addresses semantic segmentation in unstructured off-road scenes, which is important for autonomous navigation, but it is incremental as it adapts established methods rather than designing new ones.
The authors tackled the GOOSE 2D Semantic Segmentation Challenge for off-road environments by adapting an existing high-capacity pipeline with photometric distortion and exponential moving average techniques, achieving 88.8% mIoU on the validation set.
We report on the application of a high-capacity semantic segmentation pipeline to the GOOSE 2D Semantic Segmentation Challenge for unstructured off-road environments. Using a FlashInternImage-B backbone together with a UPerNet decoder, we adapt established techniques, rather than designing new ones, to the distinctive conditions of off-road scenes. Our training recipe couples strong photometric distortion augmentation (to emulate the wide lighting variations of outdoor terrain) with an Exponential Moving Average (EMA) of weights for better generalization. Using only the GOOSE training dataset, we achieve 88.8\% mIoU on the validation set.