Song, Donghui Kim, Chansu Park, Sung-Kee 2024-01-19T22:33:28Z 2024-01-19T22:33:28Z 2021-09-03 2018-06 0020-0255 https://pubs.kist.re.kr/handle/201004/121323 This paper presents a novel framework for high-level activity analysis based on late fusion using multi-independent temporal perception layers. The method allows us to handle temporal diversity of high-level activities. The framework consists of multi-temporal analysis, multi-temporal perception layers, and late fusion. We build two types of perception layers based on situation graph trees (SGT) and support vector machines (SVMs). The results obtained from the multi-temporal perception layers are fused into an activity score through a step of late fusion. To verify this approach, we apply the framework to violent events detection in visual surveillance and experiments are conducted by using three datasets: BEHAVE, NUS-HGA and some videos from YouTube that show real situations. We also compare the proposed framework with existing single-temporal frameworks. The experiments produced results with accuracy of 0.783 (SGT-based, BEHAVE), 0.702 (SVM-based, BEHAVE), 0.872 (SGT-based, NUS-HGA), and 0.699 (SGT-based, YouTube), thereby showing that using our multi-temporal approach has advantages over single-temporal methods. (C) 2018 Elsevier Inc. All rights reserved. English ELSEVIER SCIENCE INC LATE FUSION RECOGNITION BEHAVIOR A multi-temporal framework for high-level activity analysis: Violent event detection in visual surveillance Article 10.1016/j.ins.2018.02.065 1 INFORMATION SCIENCES, v.447, pp.83 - 103 INFORMATION SCIENCES 447 83 103 scie scopus 000430902400006 2-s2.0-85043476141 Computer Science, Information Systems Computer Science Article LATE FUSION RECOGNITION BEHAVIOR Computer vision Multi-temporal framework High-level activity analysis Violent event detection Late fusion Visual surveillance