Conveyor belt arrangement - Analysis and Visualization
Department of Computer Information Technology, Purdue University Northwest
The major goal of this project was to perform nalysis of conveyor belt arrangements using computer vision to identify different colored balls. Feeds include images and videos. Will be using Machine Vision to compare object arrangement seen with hundreds of thousands of stored reference images and produce results in matrices that can be exported to spreadsheets. Utilization of R to read from the spreadsheets and perform statistical report generation will be the subsequent goal of this project.
This project focuses on color classifying by K-Nearest Neighbors Classifier which is trained by R, G, B Color Histogram. It can classify White, Black, Red, Green, Blue, Orange, Yellow and Violet. If you want to classify more color or improve the accuracy you should work on the training data. You can also use the color_recognition_api to perform real-time color recognition in your projects.
Intelligent Inflation and Income Prediction System
Department of Mathematics, Statistics and Computer Science, Purdue University Northwest
Assisted a professor study economic inequality. My work focused on creating robust metrics to characterize the growing wealth disparity in the United States. Involved usage of Wide and Deep learning models to analyze and predict inflation and income levels. I led a team of six undergraduate and graduate researchers who worked to perform operations on the census data and made it available to train and test the models. A slice into our meeting minutes included me mentoring undergraduate/graduate students to adopt right dimensionality reduction techniques, etc.
Data Modelling for effective Disaster Rehabilitation
PES Institute of Technology
Applied Cloud Natural Language API hosted on GCP to build an effective model for disaster rehabilitation. This is a work that originated from my research publication - Empirical Analysis of Social Networks - which was published at the International Conference on Computer Science and Technology held in Chennai, India. (ISBN: 978-93-85225-62-8) This project was a team project comprising of Lohit Ravindra and myself. This was part of the senior design project during my undergraduate course.
CompMaps - Offline Navigation
Department of Information Science and Engineering, PES Institute of Technology
‘CompMaps’ aims at building a solution that would help find how far a person from his desired destination is without using the internet. This again, is a prototype and may not render the time exactly. However, maximum precision can be achieved in terms of directions as the whole application deals with using co-ordinates tapped from the compass that is built with integration to this application. The application performs better when there are many users who are using this application as users can share the ‘Route files’ with others. This would make the application more robust. The implementation and description of ‘Route files’ have been discussed in the further sections of the project's document.
Sentiment Analyser
CS229, Stanford University, Summer '19
This repository consists of a binary linear classifier that reads movie reviews and guesses whether they are "positive" or "negative.". The functions are described below:
1. extractWordFeatures, which takes a review (string) as input and returns a feature vector
2. learnPredictor is built using stochastic gradient descent and main functionality is to minimize the hinge loss.
3. kmeans is the function that implements k-means algorithm
