Visual Lip Reading Dataset in Turkish

This dataset consists of words and phrases commonly used in Turkish, “merhaba” (hello), “selam” (hi), “başla” (start), “bitir” (finish), “günaydın” (good morning), “teşekkür ederim” (thank you), “hoş geldiniz” (welcome), “görüşmek üzere” (see you), “özür dilerim” (sorry) and “afiyet olsun” (enjoy your meal). The dataset contains a total of 2335 instances from different resources like Turkish series, movies, songs, and vlogs. Care was taken to ensure that the word groups were evenly distributed in the dataset.

The collected full dataset contains 225 samples for the word "basla", 244 for "bitir", 268 for "merhaba", 232 for "gunaydin", 235 for "selam", 226 for "hosgeldiniz", 209 for "ozurdilerim", 224 for "gorusmekuzere", 235 for "afiyetolsun" and 237 for "tesekkurederim".

The sample dataset directory contains one instance for each class.