摘要:Automatic sign language recognition is an importanttopic within the areas of human-computerinteraction and machine learning. On the onehand, it poses a complex challenge that requiresthe intervention of various knowledge areas, suchas video processing, image processing, intelligentsystems and linguistics. On the other hand, robustrecognition of sign language could assist in thetranslation process and the integration of hearingimpairedpeople.This paper offers two main contributions: first,the creation of a database of handshapes for theArgentinian Sign Language (LSA), which is a topicthat has barely been discussed so far. Secondly,a technique for image processing, descriptor extractionand subsequent handshape classificationusing a supervised adaptation of self-organizingmaps that is called ProbSom. This technique iscompared to others in the state of the art, such asSupport Vector Machines (SVM), Random Forests,and Neural Networks.The database that was built contains 800 imageswith 16 LSA conjurations, and is a first steptowards building a comprehensive database of Argentiniansigns. The ProbSom-based neural classifier,using the proposed descriptor, achieved anaccuracy rate above 90%.