As developing economies like India are pacing towards industrial growth, the ongoing quest for increased efficiency and productivity is putting more and more pressure on operational and maintenance excellence. Industrial Automation with readiness for the Internet of Things (IoT) is set to bring about immense growth to the economy of the nation. This motivated me to pursue a unique degree in Manufacturing Processes and Automation Engineering from the prestigious Netaji Subhas Institute of Technology, University of Delhi.
During my junior year, I got involved in the development of unmanned aerial vehicle (UAV) with emphasis on real-time exploration of a search area and created a module on object recognition to identify and detect targets on the ground. The project instilled my interest in computer vision and I decided to explore it further.
In my continuous endeavor to refine my understanding of the subject, I did multiple minor projects to understand vision from different aspects as well as took related elective courses as a part of my course curriculum.
In my senior year, under the guidance of Professor Anand Gupta, I worked on analyzing and segmenting the compound document image consisting of text, figures and tables into its respective constituents. My teammate and I came up with the innovative idea of applying hierarchical contour detection along with connected component analysis to separate the tables from the rest of the constituents. The project manifested the importance of rigorous pre-processing in every application and provided me a clear understanding of the basic concepts of computer vision.
For my undergraduate thesis, I wished to apply my predilection for vision to the academic knowledge that I had acquired. I got the opportunity to work under the guidance of Professor Sachin Maheshwari towards developing model of a robot which could identify objects in the given field or arena and could backtrack to the last saved location using an obstacle avoidance algorithm. I integrated the camera module on a Raspberry Pi and accomplished real-time destination tracking using pattern recognition and backtracking.
These intriguing academic projects served as the stepping stones towards my interest in graduate studies to explore the wide gamut of computer vision and its applications.
Paper Publications and Presentation
Professor Anand encouraged me to take up the Document Image Analysis project and develop a structured approach for fast storage and retrieval of data in order to recreate the original document. My team-mate and I proposed an n-ary tree based data structure with nodes in the form of Views and a collection of Views (Layouts), which was presented in Ninth International Conference on Contemporary Computing (2016) with the paper titled CDIA-DS: A Framework for Efficient Reconstruction of Compound Document Image using Data Structure [pdf]. The research paper will be published in IEEE Conference proceedings.
EXL Analytics, one of the market leaders in analytical services, started focusing towards intelligent product solutions as the global market moved towards automation, cloud and Software as a Service (SaaS). I was thrilled with the idea of working in a product-focused environment where capabilities of artificial intelligence (AI) could be leveraged to solve real-world business problems. I joined EXL and became one of the founding members in a team that works as a focused compact start-up in larger EXL organization utilizing their business leadership and domain knowledge in the services area.
At EXL, I am primarily responsible for modelling and development of prototypes of analytical products majorly using Python. I have gained statistical knowledge and have acquired capabilities such as natural language processing and machine learning while working on a variety of products to find data-driven solutions. Towards solving a challenging problem of adaptive content extraction from unstructured scanned legal documents, we utilized my previous experience in document image analysis for the initial categorization of extracted data, which proved quite helpful in later aspects of text mining. To prevent the fraudulent claims in the healthcare industry, we achieved a significant improvement by developing an image classification tool using artificial neural networks. I have observed that the utility of advanced machine learning techniques in a product highly impacts its business area and enable organizations to work more efficiently. This feature has ignited my interest to seek diverse applications of machine learning in the industry. s
I feel excited about the integration of machine learning in computer vision as we move towards enabling deep learning in technology. The fusion can achieve notable differences in the way people interact with technology specifically in the healthcare and transportation industry. I aspire to contribute to this change that has just begun.
With the success of data-driven approach and deluge of data, intelligent automation would require effort from people in multiple domains. Along with my undergraduate majors in automation, I have had research experience in the area of computer vision and have worked on complex data problems in the industry. I am delighted to witness the opportunities that lie at the intersection of automation and vision, therefore, I wish to pursue my masters in the field of computer vision with implementation of machine learning, and aim to create a significant impact in the industry in years to come.