Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting
For analyzing the traffic anomaly within dashcam videos from the perspective of ego-vehicles, the agent should spatial-temporally localize the abnormal occasion and regions and give a semantically recounting of what happened. Most existing formulations concentrate on the former spatial-temporal aspe...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2019-11-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/19/23/5098 |
| _version_ | 1798040313728598016 |
|---|---|
| author | Rixing Zhu Jianwu Fang Hongke Xu Jianru Xue |
| author_facet | Rixing Zhu Jianwu Fang Hongke Xu Jianru Xue |
| author_sort | Rixing Zhu |
| collection | DOAJ |
| description | For analyzing the traffic anomaly within dashcam videos from the perspective of ego-vehicles, the agent should spatial-temporally localize the abnormal occasion and regions and give a semantically recounting of what happened. Most existing formulations concentrate on the former spatial-temporal aspect and mainly approach this goal by training normal pattern classifiers/regressors/dictionaries with large-scale availably labeled data. However, anomalies are context-related, and it is difficult to distinguish the margin of abnormal and normal clearly. This paper proposes a progressive unsupervised driving anomaly detection and recounting (D&R) framework. The highlights are three-fold: (1) We formulate driving anomaly D&R as a temporal-spatial-semantic (TSS) model, which achieves a coarse-to-fine focusing and generates convincing driving anomaly D&R. (2) This work contributes an unsupervised D&R without any training data while performing an effective performance. (3) We novelly introduce the traffic saliency, isolation forest, visual semantic causal relations of driving scene to effectively construct the TSS model. Extensive experiments on a driving anomaly dataset with 106 video clips (temporal-spatial-semantically labeled carefully by ourselves) demonstrate superior performance over existing techniques. |
| first_indexed | 2024-04-11T22:05:53Z |
| format | Article |
| id | doaj.art-3de66a41df2e4427a686fa4ffc20e0f0 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-04-11T22:05:53Z |
| publishDate | 2019-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-3de66a41df2e4427a686fa4ffc20e0f02022-12-22T04:00:43ZengMDPI AGSensors1424-82202019-11-011923509810.3390/s19235098s19235098Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and RecountingRixing Zhu0Jianwu Fang1Hongke Xu2Jianru Xue3School of Electronic and Control Engineering, Chang’an University, Xi’an 710064, ChinaSchool of Electronic and Control Engineering, Chang’an University, Xi’an 710064, ChinaSchool of Electronic and Control Engineering, Chang’an University, Xi’an 710064, ChinaInstitute of Artificial Intelligence and Robotics (IAIR), Xi’an Jiaotong University, Xi’an 710049, ChinaFor analyzing the traffic anomaly within dashcam videos from the perspective of ego-vehicles, the agent should spatial-temporally localize the abnormal occasion and regions and give a semantically recounting of what happened. Most existing formulations concentrate on the former spatial-temporal aspect and mainly approach this goal by training normal pattern classifiers/regressors/dictionaries with large-scale availably labeled data. However, anomalies are context-related, and it is difficult to distinguish the margin of abnormal and normal clearly. This paper proposes a progressive unsupervised driving anomaly detection and recounting (D&R) framework. The highlights are three-fold: (1) We formulate driving anomaly D&R as a temporal-spatial-semantic (TSS) model, which achieves a coarse-to-fine focusing and generates convincing driving anomaly D&R. (2) This work contributes an unsupervised D&R without any training data while performing an effective performance. (3) We novelly introduce the traffic saliency, isolation forest, visual semantic causal relations of driving scene to effectively construct the TSS model. Extensive experiments on a driving anomaly dataset with 106 video clips (temporal-spatial-semantically labeled carefully by ourselves) demonstrate superior performance over existing techniques.https://www.mdpi.com/1424-8220/19/23/5098driving anomalytemporal-spatial-semantic analysisisolation forestsemantic causal relation |
| spellingShingle | Rixing Zhu Jianwu Fang Hongke Xu Jianru Xue Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting Sensors driving anomaly temporal-spatial-semantic analysis isolation forest semantic causal relation |
| title | Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting |
| title_full | Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting |
| title_fullStr | Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting |
| title_full_unstemmed | Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting |
| title_short | Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting |
| title_sort | progressive temporal spatial semantic analysis of driving anomaly detection and recounting |
| topic | driving anomaly temporal-spatial-semantic analysis isolation forest semantic causal relation |
| url | https://www.mdpi.com/1424-8220/19/23/5098 |
| work_keys_str_mv | AT rixingzhu progressivetemporalspatialsemanticanalysisofdrivinganomalydetectionandrecounting AT jianwufang progressivetemporalspatialsemanticanalysisofdrivinganomalydetectionandrecounting AT hongkexu progressivetemporalspatialsemanticanalysisofdrivinganomalydetectionandrecounting AT jianruxue progressivetemporalspatialsemanticanalysisofdrivinganomalydetectionandrecounting |