D2-V2X benchmark and LiDAR-aligned VLM baseline raise occluded hazard recall to 24.4% and cut spatial estimation error by 77% versus zero-shot models in cooperative V2X settings.
Qlora: Efficient finetuning of quantized llms
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
D2-V2X: Depth-Driven Cooperative V2X Reasoning for Autonomous Driving
D2-V2X benchmark and LiDAR-aligned VLM baseline raise occluded hazard recall to 24.4% and cut spatial estimation error by 77% versus zero-shot models in cooperative V2X settings.