Criação de uma base de dados para o alfabeto datilológico utilizando dispositivo de interação não-convencional
Gestures are one of many ways of comunication, fundamental to people with hearing disabilities. Yet, using sign language is not as trivial, for both deaf and hearing people, resulting in dificulties to the adoption of sign language. With the objective of helping the comunication using signs, studies...
| Autor principal: | Barreto, Maísa |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2020
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/5995 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
Gestures are one of many ways of comunication, fundamental to people with hearing disabilities. Yet, using sign language is not as trivial, for both deaf and hearing people, resulting in dificulties to the adoption of sign language. With the objective of helping the comunication using signs, studies and tools using the most used methods are being developted. Among the methods utilized are the ones who use (1) cameras and sensors, (2) special gloves and hand sensors and (3) electromyographic sensors. The electromyography is a recent tecnic that is capable of identifing muscle eletric signals, witch there is no much data avaliabe specifically to sign language, apart from the others methods, as digital images that offer a great variaty of bases to represent many sign languages and alphabets. Among the objectives of this work are to prove that is interesting to use unconventional capture devices and electromyographic data to the Brazilian Sign Language retrived using the armband type device named MYO. Initialy there was the implementation of an application capable of capturing the brute data(raw data) from the device. Henceforth the implementation, two groups of volunteers were selected, fluentes and not fluentes. Once selected, these volunteers got trough data capture where they used the armband while doing the hand configurations as instructed. The data were pre processed and statistically analyzed. That said, the pre processed data underwent classification by KNN and SVM algorithmis generating the necessery data to validate the data base. The results gathered were satisfactory, in the best case were the SVM method was used, it was generated a 89% of correction, correctly classifing 578 instancies of 643 in total. With this results, its possible to conclude that the data base is valid, therefore, can be used freely. As future work, it will be investigated the possibility of using the MYO armband as a express recognition tool. |
|---|