Ved Prakash Dubey's Projects
Repository of notebooks of use-cases I worked on during my Cainvas internship at AI Technology and Systems.
An analytical and interactive approach to showcase India's effort against the ongoing COVID19 pandemic. Performed analysis of existing data using the ELK stack, keeps the user updated on live statistics, and keeps the user engaged through an interactive chatbot made with RASA which also provides them with live location based COVID19 statistics.
An accessible and robust website made using ReactJS that will perform land cover segmentation and classification from satellites and drones at the click of a button powered by powerful deep learning models served by FastAPI. We have also showed how drastically land cover changes have occurred due to environmental calamities such as thunderstorms and floods.
Multi-class emotion classification with a BERT model deployed using FastAPI
Built a binary heart disease classifier using machine learning techniques.
Resources for "Introduction to Deep Learning" course.
The wellness app
Machine Learning based feature extraction of Electrical Substations from Satellite data using Open-Source tools. Achieved test IoU of 84% using our model for semantic segmentation and a third rank on the leaderboard.
Easy Space Data Access - Making space data more accessible for people ranging from space enthusiasts to astronomers and researchers.
Converting any image with printed text into a downloadable text file. All with the help of Python and deployed with Streamlit.
Full stack machine learning project, with data visualization and analysis of the air quality of India, the effect of the pandemic on pollution, and performing Time Series Forecasting on the data using Facebook Prophet to predict future values of the AQI. Interactive and engaging website designed and the model and backend deployed on a local host using Flask.
Building a portfolio website and deploying it.
Notebook to highlight how machine learning can be used on astronomical data
Performing Pneumonia Detection and Segmentation from DICOM images. Data provided from the Kaggle competition RSNA Pneumonia Detection Challenge
In this poster, I have summarized some of the well-known loss functions widely used for Semantic Segmentation and explained where their usage can help in fast and better convergence of a model. The dataset used is a combination of the NLM Montgomery County and the Shenzhen set - Chest X-ray Database upon which we have to perform the task of binary semantic segmentation.
Voice over IP web application built using Express, Socket.IO and Node JS along with an interactive front-end.
Jupyter notebook for Jobathon 2021's problem statement