Arabic Sign Language Recognition System for Alphabets Using Machine Learning Techniques
Table 3
Arabic sign language dataset of alphabet letters.
Dataset no.
Description and samples %
Dataset 1 (training set) 2800 samples
Dataset no. 1 for the Arabic alphabet gestures for the deaf language, made with a bare hand, light background and different lights “true color” where 100 samples were taken for each letter of the Arabic alphabet
Dataset 2 (training set) 2800 samples
Dataset no. 2 for the Arabic alphabet gestures for the deaf language, made with a bare hand, dark background and different lights “true color” where 100 samples were taken for each letter of the Arabic alphabet
Dataset 3 (training set) 2800 samples
Dataset no. 3 for the Arabic alphabet gestures for the deaf language, made with a right hand and wearing gloves, white background and different lights “true color” where 50 samples were taken for each letter of the Arabic alphabet
Dataset 4 (testing set) 840 samples
Dataset no. 4 for the Arabic alphabet gestures for the deaf language, made with a bare hand “two hands” and wearing glove with different background, different angles, different lights, “true and gray image” where 50 samples were taken for each letter of the Arabic alphabet