Energy & materialsResearch findings
Development of an Efficient Vision-Language Model EM-VLM4AD for Autonomous Driving
Original published 2024.03.29 06:18 KSTAdded to Blackdeck 2026.11.14 13:31 KST
What has changed?
A multi-frame vision-language model that reduces memory usage by at least 10 times compared to existing models
Key points
- EM-VLM4AD, a lightweight vision-language model for autonomous driving, is developed to improve efficiency
- It achieves at least 10 times lower memory and FLOPs than prior models while maintaining high CIDEr and ROUGE-L scores on DriveLM
- The model extracts relevant traffic scene information based on prompts and answers various autonomous driving subtasks
Connect it to your work
Enables real-time processing in autonomous driving environments
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.