Reading Between the Lanes: Text VideoQA on the Road

TL;DR

This paper introduces RoadTextVQA, a new dataset of driving videos with questions about road signs and text, aiming to improve video question answering for driver assistance systems.

Contribution

The paper presents RoadTextVQA, a novel dataset for VideoQA focused on road sign recognition in driving videos, facilitating research in in-vehicle support and multimodal reasoning.

Findings

State-of-the-art models perform poorly on the dataset

The dataset contains 3,222 videos and 10,500 questions

Highlighting the need for improved VideoQA methods for driving scenarios

Abstract

Text and signs around roads provide crucial information for drivers, vital for safe navigation and situational awareness. Scene text recognition in motion is a challenging problem, while textual cues typically appear for a short time span, and early detection at a distance is necessary. Systems that exploit such information to assist the driver should not only extract and incorporate visual and textual cues from the video stream but also reason over time. To address this issue, we introduce RoadTextVQA, a new dataset for the task of video question answering (VideoQA) in the context of driver assistance. RoadTextVQA consists of $3,222$ driving videos collected from multiple countries, annotated with $10,500$ questions, all based on text or road signs present in the driving videos. We assess the performance of state-of-the-art video question answering models on our RoadTextVQA dataset, highlighting the significant potential for improvement in this domain and the usefulness of the dataset in advancing research on in-vehicle support systems and text-aware multimodal question answering. The dataset is available at http://cvit.iiit.ac.in/research/projects/cvit-projects/roadtextvqa