상세정보
학술대회
AI-Based Modeling Architecture to Detect Traffic Anomalies from Dashcam Videos
- 발행일
- 202210
- 출처
- International Conference on Information and Communication Technology Convergence (ICTC) 2022, pp.1480-1482
- 협약과제
- 22ZR1200, DNA 기반 국가 지능화 핵심기술 개발, 신성웅
- 초록
- This paper introduces a new modeling architecture to detect traffic anomalies using AI techniques. This study intends to reveal the effectiveness of merging extracted features which may be changed over predefined time period from dashcam video datasets. Relevant features are extracted by using a convolutional learning method and their temporal occurrence is modeled with a self-attention model. Segmented traffic accidents are classified into a couple of pre-defined groups indicating different traffic accident types. The analysis results show that the proposed modeling architecture is quite effective to identify traffic anomalies from dashcam video datasets. Additional issues for future analysis and implementations are discussed briefly as well.
- KSP 제안 키워드
- AI techniques, Attention model, Dashcam video, Learning methods, Modeling architecture, Time period, Traffic accident, Traffic anomalies