Driving with InternVL: Oustanding Champion in the Track on Driving with Language of the Autonomous Grand Challenge at CVPR 2024

Jiahan Li, Zhiqi Li, Tong Lu · 2024

Read Driving with InternVL: Oustanding Champion in the Track on Driving with Language of the Autonomous Grand Challenge at CVPR 2024 by Jiahan Li, Zhiqi Li, Tong Lu (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-07247.txt.

Description

Jiahan Li, Zhiqi Li, and Tong Lu report a winning self-driving system in a 2024 technical paper. Their entry topped the Driving with Language track of a major 2024 autonomous driving challenge. The task joins camera views of the road with language questions about the scene.

The team fine-tuned InternVL, a strong open model that handles images and text together, on the contest data set. Camera views from several angles were stitched in a fixed layout so the model could use its image skills well. A simple trick turned object centre points in the data into full boxes, adding useful training labels. Their single model scored at the top of the final leaderboard, the report states.

Catalog record

CREATORJiahan Li, Zhiqi Li, Tong Lu
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE8 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-07247

Browse the full library · About the Tower of Babel