Research papers · Posted 2024.10.14 · Publication / acceptance: author-reported
CAFuser: Condition-Aware Multimodal Fusion for Robust Semantic Perception of Driving Scenes
CAFuser: Condition-Aware Multimodal Fusion for Robust Semantic Perception of Driving Scenes
Overview
CAFuser derives an environmental-condition token from RGB images to guide multimodal sensor fusion. Modality-specific adapters share a backbone, with segmentation benchmarks examining robustness under adverse driving conditions.
Publication and version information
Publication, acceptance or venue information is author-reported on arXiv. The exact source wording is preserved below.
Author-supplied source information
Author-supplied information on arXiv.
IEEE Robotics and Automation Letters, The source code is publicly available at: https://github.com/timbroed/CAFuser
Version: arXiv v2 — manuscript revision