Global Feature Aggregation for Accident Anticipation

TL;DR

This paper introduces a novel feature aggregation method combined with LSTM to improve early accident anticipation in videos, achieving state-of-the-art results on the Street Accident dataset.

Contribution

The paper proposes a new Feature Aggregation (FA) block that refines object features considering interactions, enhancing accident anticipation performance.

Findings

Achieves earlier accident prediction by up to 0.75 seconds.

Outperforms previous methods in risk anticipation metrics.

Reports highest mean Average Precision and ATTA on the Street Accident dataset.

Abstract

Anticipation of accidents ahead of time in autonomous and non-autonomous vehicles aids in accident avoidance. In order to recognize abnormal events such as traffic accidents in a video sequence, it is important that the network takes into account interactions of objects in a given frame. We propose a novel Feature Aggregation (FA) block that refines each object's features by computing a weighted sum of the features of all objects in a frame. We use FA block along with Long Short Term Memory (LSTM) network to anticipate accidents in the video sequences. We report mean Average Precision (mAP) and Average Time-to-Accident (ATTA) on Street Accident (SA) dataset. Our proposed method achieves the highest score for risk anticipation by predicting accidents 0.32 sec and 0.75 sec earlier compared to the best results with Adaptive Loss and dynamic parameter prediction based methods respectively.