As a Tech Lead, I am responsible for design, implementation, and deployment of innovative solutions to various real-life problems related to security, surveillance, video analytics, retail, and biometric-based personal recognition systems to Reliance Jio business units for its growth in hardware/software industry. Few interesting projects are briefly mentioned below:1. Retail Checkout Automation: Developed innovative frameworks for 3D point clouds registration and RGBD- based deep net… Show more As a Tech Lead, I am responsible for design, implementation, and deployment of innovative solutions to various real-life problems related to security, surveillance, video analytics, retail, and biometric-based personal recognition systems to Reliance Jio business units for its growth in hardware/software industry. Few interesting projects are briefly mentioned below:1. Retail Checkout Automation: Developed innovative frameworks for 3D point clouds registration and RGBD- based deep net architecture for classifying retail products (with volumes) for an automated and highly accurate self-checkout system for Reliance retail stores. -Ideated and designed data (RGBD sequence) acquisition, experimental setup, and deployment scenarios
 -Novel methods for 3D point cloud registration, mesh creation, and generation of synthetic high-quality images -Novel deep net architecture using (4D convolutions) for multi-task and multi-label classification systems 2. Universal Face Detector: Developed a novel and unified architecture for fast and robust combined face detection and attribute extraction system. The envisioned framework is designed for robust(for profile faces), fast (approx. 20ms), and accurate face-based systems. -Created a data annotation pipeline for generating 1M labelled faces for training multi-task learning frameworks -Novel CNN architecture based on single-shot learning for face detection and facial attributes extraction -Simultaneous detection of the bounding box, shape, facial landmarks, age, gender, and emotion from each face.3. Other Projects: A brief description of some other designed and delivered projects include: -A novel unsupervised clustering-based approach for automatic face (photo) tagging applications. -A novel semi-supervised spark-based face recognition API particularly developed for Indian faces -Training and inference pipelines are written in python using Keras, and TensorFlow Show less

My research activities have been mainly focussed on:Biometric Systems:The focus is to investigate novel and innovative techniques, methods, and systems for biometric recognition based on hard and soft biometric characteristics. The emphasis is on designing less-constrained and user-friendly biometric technologies, which improve the quality of the human-machine interaction in unconstrained environments. In particular, his research considers technologies based on fingerprint, face… Show more My research activities have been mainly focussed on:Biometric Systems:The focus is to investigate novel and innovative techniques, methods, and systems for biometric recognition based on hard and soft biometric characteristics. The emphasis is on designing less-constrained and user-friendly biometric technologies, which improve the quality of the human-machine interaction in unconstrained environments. In particular, his research considers technologies based on fingerprint, face, voice, soft biometrics and multi-modal biometricsAmbient Intelligence: The research regarded the study of innovative less-constrained technologies able to improve the quality of the human-computer interaction and increase the applicability of biometric systems in ambient intelligence such as SmartHomes, smart offices, Smart City, Smart Airports, Automated Border Control and e-gates.Computational Intelligence:The research regarded the study of computational intelligence techniques to design an intelligent environment able to learn the specific behavior or pattern of the users, adapt and update itself during the systems operations and provide low-cost solutions to design and maintain the system. In particular, the studied computational intelligence techniques are based on Neural networks and Convolutional neural networks, Support vector machines, Fuzzy logic, Evolutionary computations, Learning theory. Show less

Worked as a Research Associate on research projects for the development of accurate and reliable multimodal biometric systems for personal authentication. The research projects were funded by Department of Science and Technology (DST), Government of India, at Biometric Research Laboratory, Electrical Engineering Department, Indian Institute of Technology, Delhi, India.In particular, I have been involved in two research projects namely:1. Biometric Based Personal Authentication… Show more Worked as a Research Associate on research projects for the development of accurate and reliable multimodal biometric systems for personal authentication. The research projects were funded by Department of Science and Technology (DST), Government of India, at Biometric Research Laboratory, Electrical Engineering Department, Indian Institute of Technology, Delhi, India.In particular, I have been involved in two research projects namely:1. Biometric Based Personal Authentication using Face Based Biometrics.The project proposed to develop face-based biometrics techniques based on infrared facial thermograms, iris, and ear biometric technologies, which can offer higher confidence/performance in the presence of imposter attack, under the grant agreement RP02309 coordinated by Department of Science and Technology (DST), India, principal investigator Prof. M. Hanmandlu, Biometric Research Laboratory, Department of Electrical Engineering. 2. Personal Authentication using Multimodal Behavioral Biometrics: Voice and Gait.The project proposed to develop behavioral biometrics based techniques based on voice and gait biometric technologies, for a personal authentication system, under the grant agreement RP02847 coordinated by Department of Science and Technology (DST), India, principal investigator Dr. B. K. Panigrahi, Biometric Research Laboratory, Department of Electrical Engineering. Show less

Inventory Control & Cost Optimization” under FGS and RMS departmentProject Details: In this project a large range (5000 approx) of data items were studied for a certain period and submitted a detailed report on classification of Inventory items on the basis of FSN (fast moving, slow moving, and non moving) analysis for FGS (Finished Goods Store) so that the limited space of inventory store can be optimized. A separate analysis has been done on the raw material items on the basis of ABC… Show more Inventory Control & Cost Optimization” under FGS and RMS departmentProject Details: In this project a large range (5000 approx) of data items were studied for a certain period and submitted a detailed report on classification of Inventory items on the basis of FSN (fast moving, slow moving, and non moving) analysis for FGS (Finished Goods Store) so that the limited space of inventory store can be optimized. A separate analysis has been done on the raw material items on the basis of ABC analysis and report is submitted to RMS (Raw Material Store) Show less