The Intuitive Supervision Model (ISM) using Convolution Neural Networks (CNN) and Unscented Kalman Filters (UKF)
Noopur Soni1, Agya Mishra2

1Noopur Soni*, Department of Electronics and Telecommunication, Jabalpur Engineering College, Jabalpur, (M.P), India. 
2Dr. Agya Mishra, Department of Electronics and Telecommunication, Jabalpur Engineering College, Jabalpur (M.P), India. 
Manuscript received on January 20, 2022. | Revised Manuscript received on January 27, 2022. | Manuscript published on January 30, 2022. | PP: 117-124 | Volume-10 Issue-5, January 2022. | Retrieval Number: 100.1/ijrte.E67820110522 | DOI: 10.35940/ijrte.E6782.0110522
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Radio frequency identification technology is one of the fastest-growing technologies in the realms of navigation, medical, robotics, communication system, logistics, security, safety, etc. Surveillance is one of the important fields where high accuracy and fast response are needed. In this research work, RFID sensors are used to track moving objects with an intelligent supervision model. The sophisticated surveillance model employs neural networks followed by an adaptive filtering technique based on an Unscented Kalman filter. A neural network is also one of the most efficient and powerful technology in the field of learning and data processing capability. A neural network has the capability of processing a mammoth amount of data because of this feature its efficiency and accuracy are quite high. This model localizes N number of objects/targets through an intelligent surveillance model, picks a random object from this pool of localized objects to track, categorizes their movement through a controlled checkpoint, and calculates the distance traveled by the moving object /target. Experimental results show that the proposed model can locate multiple-objects with the help of multiple input RFID antennas and tags and track them concerning to the RFID antennas with high accuracy and stability in the complex indoor environment and this intuitive model can be effectively implemented at the airport, railway station, shopping mall, retail management, as well as any other surveillance purpose. For this research work number of authors work, is reviewed and based on literature review this model is designed. 
Keywords: Adaptive Filtering, Convolution Neural Network (CNN), Deep Neural Network, Gauss-Newton Algorithm, Intelligent system, Indoor Positioning and tracking, Radiofrequency Identification (RFID), Unscented Kalman Filter (UKF), Received Signal Strength Indication (RSSI)
Scope of the Article: Electronics and Telecommunication