CAMPUS ABNORMAL BEHAVIOUR RECOGNITION WITH TEMPORAL SEGMENT TRANSFORMERS

Authors

  • MR. N. CHANDIRAPRAKASH, Author
  • VENKATA RAMANA Author
  • N.CHANDRAKANTH Author
  • YARRAMSHETTI KAVYA Author
  • KOTHAKOTA DHANUSH Author

Keywords:

apturing sufficient motion sequence features from videos, a novel  framework is established

Abstract

The intelligent campus surveillance system is beneficial to improve safety in school. 
Abnormal behavior recognition, a field of action recognition in computer vision, plays 
an essential role in intelligent surveillance systems. Computer vision has been actively 
applied to action recognition systems based on Convolutional Neural Networks 
(CNNs). However, capturing sufficient motion sequence features from videos remains 
a significant challenge in action recognition. This work explores the challenges of 
video-based abnormal behavior recognition on campus. In addition, a novel 
framework is established on long-range temporal video structure modeling and a 
global sparse uniform sampling strategy that divides a video into three segments of 
identical durations and uniformly samples each snippet. The proposed method 
incorporates a consensus of three temporal segment transformers (TST) that globally 
connects patches and computes self attention with joint spatiotemporal factorization. 
The proposed model is developed on the newly created campus abnormal behavior 
recognition (CABR50) dataset, which contains 50 human abnormal action classes
with an average of over 700 clips per class. Experiments show that it is feasible to 
implement abnormal behavior recognition on campus and that the proposed method is 
competitive with other peer video recognition in terms of Top-1 and Top-5
recognition accuracy. The results suggest that TST-L+ can improve campus abnormal 
behavior recognition, corresponding to Top-1 and Top-5 accuracy results of 83.57% 
and 97.16%, respectively.

Downloads

Published

25-05-2024

How to Cite

CAMPUS ABNORMAL BEHAVIOUR RECOGNITION WITH TEMPORAL SEGMENT TRANSFORMERS. (2024). International Journal of Engineering Research and Science & Technology, 20(2), 430-437. https://ijerst.org/index.php/ijerst/article/view/304