Suspicious detection in road traffic based on Kd-SLIC-K-means and SBAF-CNN using aerial view surveillance video

This paper provides a suspicious activity detection framework in road traffic based on aerial surveillance video, which combines Kd-SLIC-K-means clustering and SBAF-CNN. The suggested technique will be used to improve object segmentation and classification to give better detection results. The aerial traffic datasets are experimentally evaluated and the model has an accuracy of 98.97, which is better than the current methods. The findings show that the suggested approach can help to detect suspicious patterns in traffic situations. Nevertheless, the model has been tested under certain conditions and additional tests on large scale and diverse data sets are needed to establish its generalizability in real world scenarios.