Reliable Multi-Object Tracking Model Using Deep Learning and Energy Efficient Wireless Multimedia Sensor Networks

Abstract

Presently, sensor-cloud based environment becomes highly beneficial due to its applicability in several domains. Wireless multimedia sensor network (WMSN) is one among them, which involves a set of multimedia sensors to collect data about the deployed region. Compared to traditional object tracking models, animal tracking in WMSN is a tedious process owing to the harsh, dynamic, and energy limited sensors. This article introduces a new Reliable Multi-Object Tracking Model using Deep Learning (DL) and Energy Efficient WMSN. Initially, the fuzzy logic technique is employed to determine the cluster heads (CHs) to attain energy efficiency. Next, in the second stage, a novel tracking algorithm by the use of Recurrent Neural Network (RNN) with a tumbling effect called RNN-T is developed. The proposed RNN-T model gets executed by every sensor node and the CHs execute the tracking algorithm to track the animals. Finally, the tracking results are transmitted to the cloud server for investigation purposes. In order to assess the performance of the presented model, an extensive experimental analysis is carried out by the use of a real-time wildlife video. The obtained results ensured that the RNN-T model has achieved better performance over the compared methods in different aspects.

Keywords

• WSN
• clustering
• deep learning
• tracking
• energy efficiency
• sensor cloud