|
| Ref/year | Methods of collection dataset | Dataset size and type | Classifier | Accuracy in % |
|
| [10]/2001 | Images of bare hand | Static alphabet-Arabic manual alphabets (60 examples for each sign where it was made use of 40 training samples and 20 for testing) = 1800 samples for 30 Arabic alphabet characters | Neuro-fuzzy classifier | 93.55% |
| [11]/2001 | Colored images with glove | Static alphabet-Arabic manual alphabets | ANFIS | 95.5% |
| [12]/2005 | Colorful glove with six colored areas in different colors | Static alphabet-recognition of ArSL alphabet, number of samples is 2323 for 30 different characters for 42 different gestures with a rate of 1625 for training and 698 for testing | Polynomial classifier | 93.41% when the testing data is different from the training data and 98.41% when the training data is the same as the testing data |
| [13]/2008, 2012 | Wearing colored gloves with colored markers for the fingers and wrist | Static alphabet (30 characters) number of images (900 for training, 300 for testing) | Elman and fully recurrent networks | The accuracy rate is 89.66% when using the Elman network and when using the fully RNN the accuracy rate has improved to 95.11% |
| [14]/2010 | Free or bare hands | Videos containing sequences of characters-Arabic manual alphabet (15 characters) | (MDC)
(MLP) | 91.3%
83.7% |
| [15]/2010 | Free or bare hands | Static alphabet and number (150 images) | Euclidean distance measure based on PCA | 97.0% |
| [16]/2016 | Free or bare hands and wearing colored gloves | Static alphabet (2 datasets) (dataset 1 by bare hands 700 samples for 28 characters and dataset 2 by wearing colored gloves = 700 samples for 28 characters) | C4.5 (J48), MLP, KNN (IBK), and Naïve-Bayesian classifiers | 80.67%, 88.66%, 90.7%, 84.4% for dataset 1 and 89.5%, 94.11%, 97.5%, and 96.63% for dataset 2 |
| [17, 18]/2019 | Free or bare hands and wearing colored gloves | A dataset consisting of 54,049 gray scale images representing 32 alphabet letters for the Arabic sign language was collected by 40 signers | CNN | 97.6% |
|
