PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
February 21, 2026SensorsOpen Access

Adaptive Multiple-Attribute Scenario LoRA Merge for Robust Perception in Autonomous Driving

View Full Paper
Ask AI
Bookmark
Share

Authors

RKRyosuke KawataJLJoonho LeeYGYanlei Gu

Discussion

Loading...

Member takes

Overview

This framework enhances perception in autonomous driving across challenging environmental conditions, improving accuracy.

Key Points

  • The aim is to improve perception models for autonomous driving in complex environmental conditions.
  • Developed a parameter-efficient fine-tuning framework with dynamic LoRA experts.
  • Implemented an adaptive pipeline that selects LoRA experts based on environmental conditions.
  • Validated on semantic segmentation benchmarks such as MUSES, BDD100K, and Cityscapes.
  • Achieved up to 3.23 point improvement in mean Intersection over Union (mIoU) in single-attribute settings.
  • In multiple-attribute scenarios, merged LoRA experts outperformed the baseline by up to 5.99 points.
  • Demonstrated effective generalization capabilities in data-scarce conditions.

Cite This Study

Kawata et al. (2026) studied this question.

synapsesocial.com/papers/69994bef873532290d020125https://doi.org/10.3390/s26041336
View Full Paper
Ask AI
Bookmark
Share