Dataset Information

Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting.

ABSTRACT: 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.

SUBMITTER: Zhu R

PROVIDER: S-EPMC6928745 | biostudies-literature | 2019 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting.

Zhu Rixing R Fang Jianwu J Xu Hongke H Xue Jianru J

Sensors (Basel, Switzerland) 20191121 23

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 ...[more]

PMID: 31766458

Similar Datasets

Project description:Grey matter (GM) atrophy occurs early in primary progressive MS (PPMS), but it is unknown whether its progression involves different brain regions at different rates, as is seen in other neurodegenerative diseases. We aimed to investigate the temporal and regional evolution of GM volume loss over 5years and its relationship with disability progression in early PPMS. We studied 36 patients with PPMS within five years from onset and 19 age and gender-matched healthy controls with clinical and imaging assessments at study entry and yearly for 3years and then at 5years. Patients were scored on the expanded disability status scale (EDSS) and MS Functional Composite (MSFC) at each time-point. An unbiased longitudinal voxel-based morphometry approach, based on high-dimensional spatial alignment within-subject, was applied to the serial imaging data. The rate of local (voxel-wise) volume change per year was compared between groups and its relationship with clinical outcomes was assessed. Patients deteriorated significantly during the five years follow-up. Patients showed a greater decline of GM volume (p<0.05, FWE-corrected) bilaterally in the cingulate cortex, thalamus, putamen, precentral gyrus, insula and cerebellum when compared to healthy controls over five years, although the rate of volume loss varied across the brain, and was the fastest in the cingulate cortex. Significant (p<0.05, FWE-corrected) volume loss was detected in the left insula, left precuneus, and right cingulate cortex in patients at three years, as compared to baseline, whilst the bilateral putamen and the left superior temporal gyrus showed volume loss at five years. In patients, there was a relationship between a higher rate of volume loss in the bilateral cingulate cortex and greater clinical disability, as measured by the MSFC, at five years (Pearson's r=0.49, p=0.003). Longitudinal VBM demonstrated that the progression of GM atrophy in PPMS occurs at different rates in different regions across the brain. The involvement of the cingulate cortex occurs early in the disease course, continues at a steady rate throughout the follow-up period and is associated with patient outcome. These findings provide new insights into the characteristics of GM atrophy across the brain in MS, and have potential consequences for the selection of brain atrophy as an outcome measure in neuroprotective clinical trials.

Project description:Food consumption is closely associated with resource consumption and environmental sustainability. An unreasonable dietary pattern would cause great pressure or damage to resources and the environment. It is particularly important to reduce the negative impact of household food consumption on resources and the environment while simultaneously ensuring people's nutrient intake and health. This study applied the China Health and Nutrition Survey (CHNS) database to quantitatively study the spatial-temporal analysis of multiple footprints of household food consumption at multiple scales and explored the driving mechanism of the multiple footprints. The results showed that, except land footprint (LF), the other four types of footprints all decreased at varying degrees; the water footprint (WF), carbon footprint (CF), nitrogen footprint (NF) and energy footprint (EF) decreased by 18.24%, 17.82%, 12.03% and 20.36%, respectively, from 2000 to 2011; multiple footprints of food consumption of household in Guizhou was the highest among the 12 provinces involved in the study; this shows that resource consumption (water, energy and land resource) and environmental influences (CO2 emissions and nitrogen emissions) brought by food consumption of per household in Guizhou are much greater than in other provinces, which has a negative influence on sustainable development; by analyzing the driving factors of multiple footprints, it is shown that nutrient intake, household attributes, educational level and health conditions were significantly correlated to multiple footprints. Among them, nutrient intake has greater impact on the multiple footprints of Chinese household food consumption. By comparing multiple footprints of different dietary patterns, it was found that the current Chinese dietary pattern would cause excessive resource consumption, which would bring more pressure on resources and the environment. Adjusting household living habits would possibly reverse the unsustainable situation, such as reducing the consumption of animal-derived foods and adjusting the dietary pattern of households with a higher educational level and income status. Chinese Dietary Guidelines 2016 has better sustainability; the promotion of this dietary pattern across the country would help China to relieve the pressure on resources and environment from the consumer side, promoting the realization of sustainable development.

Dataset Information

Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting.

Publications

Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets