This repository is part of Vilnius University sowftare engineering bachelors thesis made by Tomas Kondrotas. The repository contains a complete Server and Client website made for lithuanian sign language recognition data set gathering as well as jupyter notebooks that were used to train models for sign classification.
Web application server written in ASP.NET
UI application written in Blazor
Jupyter notebooks containing 3 different training approaches: Hybrid CNN + LSTM, vanilla ViViT, pretrained ViViT.