Key Projects

Number Plate Detection and Reading

• Devised normal ANPR model by creating a pipeline using two YOLO Models in Pytorch and Darknet
• Number Plate Detection Model called in Character Reading Model, former passed cropped images to latter
• Labeled and Trained the model over 1000+ dataset in each by streamlined workflow provided by Roboflow
• Achieved mAP(mean avg precision) of 0.993 in Number Plate Detection Model using YOLOv5 in Pytorch

Online Shopping System

• Scripted in PHP using XAMPP as virtual server & contains admin and user side similar to ecommerce website

AI Virtual Mouse

• Created and AI-based Mouse Controller by detecting Hand Landmarks and then tracking and clicking based on these points, also applied some smoothing techniques to make it practically more usable and reliable

Tweet Emotion Recognition

• Recurrent neural network (RNN) devised for training over 15k+ tweets, each tweet classified in one of 6 emotions
• Tokenizer used for Padding & Truncating Sequenes and performed NLP tasks using tensorflow

Decision Review System

• Designed DRS using tkinter module to build GUI application and openCV for detection & tracking purposes

To-Do-List

• User Specific registration via email & password using MongoDB service for encryption and OAuth for sign-in
• Developed client-side(front-end) using ReactJS and server-side(back-end) using NodeJS and ExpressJS