Vasileios Kochliaridis

Computer Vision Algorithms

Implementations of computer vision algorithms, including real-time adaptive thresholding with an integral image, sharpening with kernels, webcam sketch converter and and more!

Back

Traffic Sign Detector

This is a state-of-the-art Traffic Sign detector. It uses an "Inception-Resnet-v2" CNN to classify Chinese traffic signs with a high accuracy.

Back

Email Spam Detector

My first Machine-Learning project! I was tired of getting emails in my spam folder, so I decided to implement a custom Neural Network using the Tensorflow framework that would accurately detect spam emails.

Back

Big Data Algorithms

Having to deal with big data is too difficult. Not only, normal algorithms need a lot of time to be applied in large amounts of data, but they also require a lot of memory (RAM) to run. I've implemented a toolbox of efficient big data algorithms that handle both memory and execution time.

Back

Deep Reinforcement Learning Agents

AI is capable of performing as good as humans in games! They just need a little bit of training. In this Project, I've implemented a C51 agent (Distributional Reinforcement Learning) that can master any game in short time. This agent is not as fast as Google's Rainbow agent, however the algorithm converges relatively fast.

Back

Fast Chat

Tired of using communication platforms (such as Facebook, Messenger, Instagram, Skype, Viber, etc) that store personal data? You could try Fast Chat for Desktop! It's a simple chat application for You and Your team that stores no information at all. Also, It can run on multiple operating systems (such as Windows, Linux, Mac, etc).

Back

Java Design Patterns

Implementations of common design patterns in Java.

Back

MovieRadar

Implementations of a custom local search engine using Lucene API. The application downloads data from TMDB (Movies, Reviews, Actors, etc). Then, Lucene is used to store and index the data. The application uses multi-threading and JsonIter API for fast Json page parsing. Combining all these APIs, the application is able to download, parse and index large amount of data in seconds.

Back

Autonomous Vehicles

Implementations of an agent that can navigate fast & safe a Self-Driving vehicle. The agent was trained using the Deep Reinforcement Learning algorithms with the tf-agents library in Python. The simulation is done in CARLA Simulator.

Back