Project: Thesis

A Python project that was developed as a university assignment for the subject of Image Processing. The program takes an input image and a reference dataset of photos. The goal is to colorize the greyscale image using a trained support vector machine. To achieve that, we have implemented a variety of image processing techniques. First, we change color spaces from RGB to LAB. Then, we apply the SLIC algorithm to find the group of superpixels for each image. These segments along with SURF and GABOR features are given as input for the SVM. Using scikit-learn, we use machine learning techniques to predict the color of a superpixel using the dataset superpixels as reference. The output of the program returns the colorized version of the input image.

A Python project that was developed as a university assignment for the subject of Pattern Recognition. The program uses the "MovieLens 100K" dataset that includes movie ratings of random users. The data included is processed using the Pandas library. We implement three algorithms: Basic Sequential Algorithmic Scheme (BSAS), K-means and hierarchical clustering. Finally, we use a neural network with multilayer perceptron and least squares linear regression to make predictions on movie ratings.

A Python project that was developed as a university assignment for the subject of Bioinformatics. The goal of this assignment was to process hypothetical sequences of nucleotides on seven different scenarios. The Biopython library is used for biological computation. The Needleman-Wunsch algorithm is utilized for sequence alignment along with Hidden Markov Models with the Viterbi algorithm.

A Python project that was developed as a university assignment for the subject of Multimedia Systems. The program first implements Run-length encoding (RLE) for a sample image of Mona Lisa. Then, Differential pulse-code modulation encoding (DPCM) is applied on a sample TV advert. We perform motion prediction for that advert and object removal on a video scene that features moving characters on a stage. For the error frame calculations, we use the sum of absolute differences (SAD) metric and macroblock techniques.

A Python project that was developed as a university assignment for the subject of Code and Information Theory. The program features a client-server implementation using sockets. It simulates a simple exchange of an image between the client and the server. Once the user inputs a specific parameter on how the image should be compressed, the Shannon-Fano coding is applied. The image is compressed and sent from the client to the server, which then applies decompression to reconstruct the image.