HARSH KAPADIA
@nirmauni.ac.in
Assistant Professor, Electronics and Instrumentation Engineering Department
Institute of Technology, Nirma University
Prof Harsh K Kapadia is working as an Assistant Professor in Electronics and Instrumentation Engineering Department. He has more than 10 years of teaching and research experience. He has obtained MTech in Instrumentation and Control Engineering from Dharmsinh Desai University, Nadiad in 2012. He has published more than 15 research articles in national/international conferences. He has guided 7 post graduate students and several graduate students for their project work. He has filed two Indian patent along with other professors of the institute. He has successfully carried out three minor research projects funded by Nirma University. He is pursuing PhD on "Applications of advanced imaging techniques for concrete damage characterization". He has mentored Team Con-Sol-E for Mitsubishi Electric Cup ( A national level automation competition for students) which won the silver cup consecutively three times in a row (2017,2018 and 2019) and gold cup in 2020. His research areas include Image Pr
EDUCATION
Master of Technology in Instrumentation and Control Engineering
RESEARCH INTERESTS
Machine Vision Image Processing Embedded Systems
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
Harsh Kapadia, Paresh V. Patel, Jignesh B. Patel, and Nikhil Kanani
Springer Science and Business Media LLC
Devansh Rana, Simran Lulla, Harsh Kapadia, and Ankure Dwivedi
IEEE
Water being the most important for existence of life and hence there exists a need to monitor it on a regular bases with the rising need. In our experimentation setup efforts have been made to minimize the unknowns and provide a standard as per the basic laws and provide a suitable environment to obtain repeatable readings. Light of the incident wavelength can affect the excitation of electrons in the atomic or molecular ground state to higher energy levels, giving rise to an absorbance at wavelengths specific to each molecule. The purpose of this study is to provide better insights to actual deployment of sensing systems alongside further development in the domain where existing piping infrastructure is disturbed to a minimum. Sensing system developed aims to nullify the existing shortcomings and put forward a solution to inline contactless TDS measurement on a repeatable basis by using machine learning based computations. This will provide added advantages like very low maintenance cost, specific elemental analysis, and easy integration. Further discussion on challenges, and improvements for a wider application scope is discussed.
Darsh Patel, Hitika Dalwadi, Hetvi Patel, Prasham Soni, Yash Battul, and Harsh Kapadia
Springer Nature Singapore
Harsh K Kapadia, Paresh V Patel, and Jignesh B Patel
Deanship of Scientific Research
Harsh Kapadia, Nimit Soneji, Anirudha Rotti, Paresh V. Patel, and Jignesh B. Patel
Springer International Publishing
Pushpit Jain, Rishi Hiran, Harsh Kapadia, Paresh Patel, and Jignesh B. Patel
IEEE
There is a drastic increase in the usage of Artificial Intelligence, Machine Learning, and Deep Learning over the past decade, and innovative applications using these technologies are being developed. The development of such applications requires a tremendous amount of data and high computational power for training and deployment. Though cloud computing is a good way for development with the increasing amount of data, cloud computing proves out to be costly and slow. Edge computing devices have become the need of the hour, which could provide high computational power locally, saving on the costs of communicating with the server and providing a faster way of processing. The issue is addressed in the presented work with the use of commercially available edge computing devices. The developed method is tested for efficiency and computation speed. For the case study, the application of concrete crack detection using a convolutional neural network is considered. Transfer learning-based approach is adopted with the use of a pre-trained inception v3 CNN model. A concrete crack dataset is prepared for training and testing the model. In order to prove the computational efficiency of edge computing, two separated models were trained ad tested. A CPU standalone model and another model compatible with CPU and edge computing device were trained and tested. The experimental results show that the CPU with edge computing device requires much less computational time as compared to the CPU standalone mode and is at least 100 times faster in computation.
Harsh Kapadia, Alpesh Patel, Jignesh Patel, Shivam Patidar, Yash Richhriya, Darpan Trivedi, Priyank Patel, and Meet Mehta
IEEE
Municipal solid waste management has been one of the most critical issues of urban cities today. Increasing population, constructions, industries, etc. are the major factors creating a large amount of waste that is dumped onto the landfill sites. Various systems have been proposed and are under the utilization for the management of municipal waste which includes mechanical vibration-based size-based sorters, eddy current sensor-based sorting of metallic waste, automatic optical waste sorters, etc. This paper focuses on a novel solution for solid waste segregation using the concepts of machine vision and deep learning. The proposed concept is tested for the segregation of solid dry waste particularly plastic bottles, aluminum cans, and tetra packs. The prototype system developed for the segregations works at high speed and accuracy. The prototype system sorts 250 objects per minute with an average accuracy of 96%. The proposed novel idea be extended and implemented for other types of waste segregation and can include more categories of solid dry waste. The system provides a solution for the ever-challenging municipal waste management problem.
Harsh Kapadia, Hardik Brahmbhatt, Yuvrajsinh Dabhi, and Sajan Chourasia
Elsevier BV
Harsh Kapadia, Ripal Patel, Yash Shah, J. B. Patel, and P. V. Patel
Springer International Publishing
Vaishali Mishra, Harsh K. Kapadia, Tanish H. Zaveri, and Bhanu Prasad Pinnamaneni
Springer Singapore
Alpesh Patel, Rohit Singh, Jignesh Patel, and Harsh Kapadia
Springer International Publishing
Kumari Jyotsna, Ankit Sharma, and Harsh Kapadia
IEEE
This paper deals with the technical details involved in interfacing SCADA with PLC in Wind Technology. It discusses the factors responsible for multi source and multi channel data recorder by remote Data Acquisition feature (Database generation) available in InduSoft SCADA that can record the data near the source of measurement. This software has all most all driver connectivity facility which will help to configure my PLC with InduSoft SCADA, by which data can transfer remotely. A personal computer system performs all duties as programming and data display. Also all channels will be synchronized and data acquisition occurred simultaneously.
Priyabrata Majhi, Harsh K Kapadia, Alpesh I Patel, Jayesh J Barve, and Divyesh R Raninga
IEEE
The main motive of this paper is to implement, analyze and compare few standard closed loops PID tuning methods and to develop auto tuning for single board heater system (SBHS) using relay feedback method. For system identification, step test method is used to find out system parameters. Widely accepted PID controller is used to control the temperature of the system with the PID controller parameter values obtained from Zeigler Nichol's tuning method. Better to this conventional ZN, Relay Feedback tuning based PID controller is used, whose main idea was proposed by Amstrom-Hagglund in 1984. This paper also includes implementation of LQR(linear quadratic regulator) controller over linear system i.e. SBHS to reject disturbances with set point tracking, number of analysis and results has been taken for both LQR and PID controller.
Sapan Thakker and Harsh Kapadia
IEEE
This paper presents innovative approach towards design, developing, and implementation of image processing based application using embedded vision platform. Android provides platform for developing embedded system. By using OpenCV library, image processing based algorithm can be implemented on android devices. By using OpenCV with android embedded vision based system can be developed which can be replaced for machine vision based system. Android based embedded vision system reduces size of the system and also cost effective solution for industries. The android application presented in this paper performs basic operations like colour transforms, edge detection, morphological operation etc. The Paper also describes the flow of developing embedded vision applications using android devices along with some problems and its solutions.
Jigar Senjalia, Parinda Pandya, and Harsh Kapadia
IEEE
In this paper, we propose method to easily conduct the wheel alignment measurement by using Camera Calibration and Laser Triangulation. By taking advantage of laser projection which is ubiquitous in structured environments, our method makes it convenient for a mobile operation to collect the data needed for calibration. The proposed method estimates the relative position of the wheel by first determining the pose of each plane of the wheel and by comparing to standard data; we can easily find the misalignment of the wheel. The algorithm is validated first by performing experiments on simulated data and investigating its sensitivity for measurement to noise.
Parinda Pandya, Jigar Senjalia, and Harsh Kapadia
IEEE
Object recognition is one of the problems in computer vision and so many techniques have come up to solve. All of them employ machine learning, because the computer has to learn first and use it in future to say whether the query image matched or not. These proposes approaches for object recognition by applying scale and rotation invariant feature transform in an automatic segmentation algorithms like FAST, SURF, SIFT, ORB etc. The features should be discrete and stable so that it can be used for matching an object in different views. At first, an object is trained to find best features. The object can be recognized in the other images by using achieved feature points. The results should show that the proposed approach is reliable for object detection and should be robust to the noise.
Kinjal A. Shah, Harsh K. Kapadia, Vipul A. Shah, and Maurya N. Shah
IEEE
This paper discusses about localization of license plate from vehicle images using different methods. And main focus is on Mean Shift Algorithm which is a non parametric feature space analysis technique. It defines a window around each data points and shifts the center of the window to the mean of the data point though a Mean-Shift Vector till it converges and window shifts to the denser region. The other methods used are Crop Object, Morphological operations, Localize the object by thresholding and area of object. We also discuss how Mean-Shift algorithm is accurate enough and robust to be used in the same area. This includes the basic definition of the algorithm and mathematical equations.