Comparison of Collective Diverse Arabic Sign Language Dataset

Machine learning researchers from all around the world continue to work out the best ways to collect data efficiently. Data collecting has recently emerged as a key concern for two primary reasons. Despite the fact that machine learning is making significant progress, there may not be enough labelled data for some new applications. Furthermore, deep learning methods have the benefit of automatically creating features, which is not the case with traditional machine learning methods. With this, model design becomes more affordable, although more labelled data may be required. Particularly, the collection of data research has been on the increase in recent years including data management, computer vision, machine learning, and natural language processing. The main reason for this is that it's necessary to handle and process enormous quantities of data successfully. This study primarily aims to present a publicly available dataset comparison that includes large samples of Arabic sign language images for the goal of sign language classification. This dataset collection has a lot of different videos and images that show different moves. The primary objective of this comparison is to show that there are different types of sign language datasets such as words based sign and alphabet based sign furthermore, the comparison include the Background of the Datasets, the Size of the Datasets, the Number of Samples, the Number of Training, Testing, and Validation Samples, the dataset types, and RGB or Binary Images. The main goals of future study will be to improve the method and test the model using AASL-annotated data.