Conducted research on anomaly detection in textured surfaces using following approaches.– Weakly supervised anomaly detection using CNNs.– Semi-supervised anomaly detection using Autoencoders and GANs.– Supervised anomaly detection using CNNs.AnoNet: Weakly Supervised Anomaly Detection in Textured Surfaces Using CNNs•Developed a CNN architecture capable of learning to detect the actual shape of anomalies in textured surfaces from weakly labelled data. •Proposed a… Show more Conducted research on anomaly detection in textured surfaces using following approaches.– Weakly supervised anomaly detection using CNNs.– Semi-supervised anomaly detection using Autoencoders and GANs.– Supervised anomaly detection using CNNs.AnoNet: Weakly Supervised Anomaly Detection in Textured Surfaces Using CNNs•Developed a CNN architecture capable of learning to detect the actual shape of anomalies in textured surfaces from weakly labelled data. •Proposed a filter bank based initialization technique leading to stellar performance. AnoNet achieves state of the art performance with an average F1 score of 0.91 and AUROC value of 0.92 across four challenging data-sets.•AnoNet is an extremely compact network with just 65k parameters. It generalizes well to similar anomaly detection tasks and avoids over-fit to the weakly annotated data. It can learn to detect anomalies from limited number of images (as few as 53 images).Road Crack Detection and Segmentation using CNNs•Created a dataset of normal and crack images from weak annotation and built a model using transfer learning to detect cracks in top-down road images for the Ministry of Transportation of Ontario (MTO). •Achieved an accuracy of 96.6% on the validation set of 440 images and used a sliding window approach to perform patch based crack segmentation.Automatic Redaction of Video Recordings using Deep Learning•A deep learning system using MTCNN and ResNet50 trained on VGGFace2 data-set for the task of automated redaction of video recordings.•All the faces except the person(s) of interest are redacted from the video. Show less

SyDE 252 - Linear Systems and Signals: Modelling and analysis of linear systems. Discrete time systems, continuous time systems; difference and differential equations; impulse and frequency response. Complex frequency, functions of complex variables, transform domain techniques: Z transforms; Fourier analysis, Laplace transform. Transfer functions and frequency response, frequency domain analysis of linear systems; sampling theory, stability, and linear filters. Conducted weekly… Show more SyDE 252 - Linear Systems and Signals: Modelling and analysis of linear systems. Discrete time systems, continuous time systems; difference and differential equations; impulse and frequency response. Complex frequency, functions of complex variables, transform domain techniques: Z transforms; Fourier analysis, Laplace transform. Transfer functions and frequency response, frequency domain analysis of linear systems; sampling theory, stability, and linear filters. Conducted weekly Tutorial sessions for a class of ~80 students. Show less

Handled sales of Diagnostic Imaging solutions namely MRI, CT Scanners, Angiography, Molecular Imaging and Radiography Systems for the government medical colleges and hospitals for western region of India ( comprising of the states of Maharashtra, Madhya Pradesh, Goa, Chhattisgarh and Gujarat.Increased the overall business volume of the region by ~30% with an increment of INR 78 Million ( ~ USD 1.2 Million) in FY17 as compared to FY16.

Learnt embedded c, Arduino programming, AVR programming, sensor interfacing. Successfully implemented those ideas and made working models based on the same concepts.

I was a part of the IETE-ETSA student council at Fr.CRIT,Vashi. We organized various events throughout the entire year. Ranging from interview emulation event called Judgement Day to Robotics. Organizing those events was a fun filled experience. It thought me quite a lot of things like how to manage events, how to coordinate people, how to settle disputes and many more.