PROJECTS
Latest Projects
Classification of Tissues in Zebrafish
Masters Thesis
Working with high quality images obtained from Hershey hospital. Investigating Image Segmentation techniques to segment out the tissue from the rest of the images. Also investigating classification techniques to help in solving the problem.
Interactive games in Python
Python Project
Made various games like Pong, Memory Game, Stopwatch, etc using Python. It was part of an online course on coursera to learn Python. Successfully completed the course and familiarized myself with Python.
Classification of Erythemato-Squamous diseases
Pattern Recognition and Machine Learning Project
Worked with an unbalanced dataset to solve a multiclass classification problem. Feature ranking, feature selection, bagging and a variety of classifiers were used for prediction. 94% accuracy was obtained in correctly classifying the diseases. MATLAB was used for implementation.
Survey of Face Recognition Techniques
Image Processing Project
Implemented state-of-the-art algorithms like Eigenfaces, Fisherfaces etc used for face recognition. Multiple databases were used to compare the accuracy of all methods. Comparison amongst all the techniques was drawn based on accuracy, complexity, time and cost efficiency and databases. User-friendly GUI was also implemented.
Visual Categorization with Bag of Keypoints
Computer Vision Project
Solved the problem of object recognition using K-means algorithm and Naive Bayes in MATLAB. Combination of Harris Affine detector, SIFT features and recognition algorithms was used to correctly classify objects. 2 databases having multiple objects in each was used. Easy to use GUI was developed.
Feature Extraction and Matching using Harris Corners and SIFT Interest Points
Computer Vision Project
Computed an algorithm for detecting corners and features in images, matching feature points amongst them to compute matchability and repeatability scores to show how closely related the images are. Performed Image Warping using the above matched features in MATLAB.
Face Detection and Recognition for Security and Surveillance Systems
Final Year B.E. Project
Designed a cost and time efficient system to provide real time multiple face recognition using MATLAB. Used Viola - Jones algorithm and SURF features to compare the query image along with the existing database for recognition. GUI with various functionalities was created.
Bond Market Prediction using an Ensemble of Neural Networks
Technical Paper - http://www.ijcaonline.org/archives/volume82/number4/14105-2144
Designed an ensemble of neural networks including Back Propagation network, SVM and Linear Regression model for prediction over a 5 and 10 day period using MATLAB.