Swarali's Projects
AWS Face Recognition as a Service(PaaS)
A real - time face recognition application which is serverless and uses Function as a Service model of cloud computing.
Raspberry Pi is used as the edge device to record videos and AWS Lambda function based on a container image(deep learning model) is used to provide the face recognition service.
Python Docker PyTorch AWS Lambda AWS S3 AWS DynamoDB AWS API Gateway AWS Raspberry Pi
Movie Recommendation Engine
The entertainment industry today is massively driven by Movies.Ever since the emergence of on-demand platforms like Netflix, Prime-Video, etc. the need to leverage data and technology to streamline the process of content consumption has increased manifold.
With the advent of Machine Learning, these platforms can accurately recommend movies to watch based on our watching habits.These models use clustering techniques to identify data points to make accurate predictions.
In this project, we aim to compare such models using Collaborative Filtering and Matrix Factorization and Item-based KNN, and test it on the 20M MovieLens dataset to predict users' movies to watch.
JavaScript Jupyter Notebook CSS HTML Python Perl Shell
AWS Face Recognition as a Service (IaaS)
Developed an elastic cloud application that provides image recognition service to the users by using IaaS resources offered by AWS.
Implemented the application to handle multiple user requests concurrently and be capable of scale-in and scale-out based on the current demand.
Implemented the whole framework in a multi-tiered architecture developed using React JS on the frontend and Java SpringBoot framework for our REST application.
Python AWS S3 AWS SQS AWS EC2 Java Spring Boot
Continuous Glucose monitoring system
Developed and trained a machine learning model on various machine learning algorithms viz. Logistic Regression, Random Forest, GaussianNB, SVM, K-Nearest Neighbor to predict the timing of insulin ingestion by guessing meal intake from Continuous Glucose Monitor’s data.
SARIMA RNN Kalman
SmartHome - Gesture Control Application
System to watch and practice SmartHome gesture videos. Functionality to watch, record and send videos to cloud server to store using Restful API.
The user is shown a video of a gesture.
The user can replay the video at least 3 times.
Upon clicking the “PRACTICE” button, the user can capture his or her own video through the smartphone’s front camera for a period of at most 5 seconds.
The videos are uploaded to a server.
Java Android Studio
Twitter Emotion Recognition
With the use of Recurrent Neural Networks, a model is created and trained to learn to recognize emotions in tweets.
The dataset has thousands of tweets each classified in one of 6 emotions – love, fear, joy, sadness, surprise and anger.
RNN Jupyter Notebook Keras Tensorflow
COVID-19 Symptom Checker
Developed an Android App that measures Heart-Rate, Respiratory-Rate (Using sensor values and Peak detection Algorithm) and collects COVID-19 related symptoms and stores them in a database in the smartphone
Android App that collects COVID-19 related symptoms and stores them in a database in the smartphone The app has 2 pages.
In the first page it presents the user with two sign measurement techniques: a) heart rate sensing, and b) respiratory rate sensing.
Java Android Studio
Sign Language Conversion For Tiny-Tots
Built an educational module to teach English grammar and spellings to physically challenged children by training a dataset of 10000+ images using CNN.
Developing a testing platform wherein students displayed the spelling of the word displayed on the screen.
The camera captured the images and then classified it using the pre-trained network.
Based on the classification the results were displayed.
Matlab Deep Learning Convolutional Neural Networks Hand Gesture Recognition HTML CSS Javascript