Transfer-LMR: Heavy-Tail Driving Behavior Recognition in Diverse Traffic Scenarios

TL;DR

Transfer-LMR is a modular training method that significantly improves recognition of rare and underrepresented driving behaviors in diverse traffic scenarios, addressing the heavy-tail distribution challenge.

Contribution

The paper introduces Transfer-LMR, a novel training routine that enhances recognition performance across all driving behavior classes, especially rare behaviors, in heavy-tailed datasets.

Findings

Improves recognition accuracy for rare driving behaviors.

Effective across diverse traffic scenarios and datasets.

Addresses heavy-tail distribution challenges in behavior recognition.

Abstract

Recognizing driving behaviors is important for downstream tasks such as reasoning, planning, and navigation. Existing video recognition approaches work well for common behaviors (e.g. "drive straight", "brake", "turn left/right"). However, the performance is sub-par for underrepresented/rare behaviors typically found in tail of the behavior class distribution. To address this shortcoming, we propose Transfer-LMR, a modular training routine for improving the recognition performance across all driving behavior classes. We extensively evaluate our approach on METEOR and HDD datasets that contain rich yet heavy-tailed distribution of driving behaviors and span diverse traffic scenarios. The experimental results demonstrate the efficacy of our approach, especially for recognizing underrepresented/rare driving behaviors.