AccidentBlip: Agent of Accident Warning based on MA-former

2024-04-18

AI Summary

Existing vision-only accident prediction methods suffer from low accuracy and high computational cost in complex traffic scenarios—particularly those involving vehicle-to-vehicle (V2V) and vehicle-to-everything (V2X) interactions—and lack native support for multi-camera collaborative modeling. This paper introduces AccidentBlip, the first end-to-end framework for multi-view vision-only accident early warning. Its core is the novel Motion Accident Transformer (MA-former), which features: (1) temporal attention—replacing standard self-attention—to effectively capture long-range motion patterns; (2) cross-frame query residual connections to enhance dynamic feature propagation over time; and (3) multi-camera query concatenation to jointly encode spatiotemporal interactions among vehicles and infrastructure. Evaluated on the DeepAccident benchmark, AccidentBlip achieves state-of-the-art performance on both accident detection and prediction tasks, with particularly significant gains in V2V and V2X sub-scenarios over prior best methods.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Security and Privacy: Large-scale security measurements
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

In complex transportation systems, accurately sensing the surrounding environment and predicting the risk of potential accidents is crucial. Most existing accident prediction methods are based on temporal neural networks, such as RNN and LSTM. Recent multimodal fusion approaches improve vehicle localization through 3D target detection and assess potential risks by calculating inter-vehicle distances. However, these temporal networks and multimodal fusion methods suffer from limited detection robustness and high economic costs. To address these challenges, we propose AccidentBlip, a vision-only framework that employs our self-designed Motion Accident Transformer (MA-former) to process each frame of video. Unlike conventional self-attention mechanisms, MA-former replaces Q-former's self-attention with temporal attention, allowing the query corresponding to the previous frame to generate the query input for the next frame. Additionally, we introduce a residual module connection between queries of consecutive frames to enhance the model's temporal processing capabilities. For complex V2V and V2X scenarios, AccidentBlip adapts by concatenating queries from multiple cameras, effectively capturing spatial and temporal relationships. In particular, AccidentBlip achieves SOTA performance in both accident detection and prediction tasks on the DeepAccident dataset. It also outperforms current SOTA methods in V2V and V2X scenarios, demonstrating a superior capability to understand complex real-world environments.

Problem

• Accident Prediction
• Complex Traffic Scenarios
• Multi-camera Synchronization

Innovation

• AccidentBlip
• MA-former
• time understanding assistant