Artificial Intelligence Based Real Time Surveillance and Object localization System

Real-time (RT) surveillance and object localization are very important for the detection of objects and their activities. It helps society to monitor the anomalous behaviours of objects and enhance security systems. For accurate and precise detection, an effective, fast, and reliable algorithm is needed. This paper proposes a novel real-time technique of surveillance and object localization, i.e., object and location prediction. The first step is to propose a novel technique of surveillance, in which the encoder-based features are extracted and then used for the prediction of object locations and their classes. The proposed technique is trained with a standard dataset. After that, in the second step, the proposed technique is deployed on a Jetson-Nano ARM Cortex-A57 based developer board for RT implementation. The proposed methodology shows the effective results in RT with an approximate accuracy of 97% and 99% with standard datasets.