Visually impaired (VI) people face a set of challenges when trying to orient and contextualize themselves. Computer vision and mobile devices can be valuable tools to help them improve their quality of life. This work presents a tool based on computer vision and image recognition to assist VI people to better contextualize themselves indoors. The tool works as follows: user takes a picture $\rho$ using a mobile application; ρ is sent to the server; ρ is compared to a database of previously taken pictures; server returns metadata of the database image that is most similar to ρ; finally the mobile application gives an audio feedback based on the received metadata. Similarity test among database images and $\rho$ is based on the search of nearest neighbors in key points extracted from the images by SIFT descriptors. Three experiments are presented to support the feasibility of the tool. We believe our solution is a low cost, convenient approach that can leverage existing IT infrastructure, e.g. wireless networks, and does not require any physical adaptation in the environment where it will be used.
Técnicas de Deep learning vêm mostrando avanços em várias tarefas de aprendizado de máquina. Porém a implementação dessas técnicas é muito complexa. Assim, para ajudar na implementação de projetos de Deep Learning, plataformas estão sendo criados. Já existe uma quantidade considerável destas plataformas disponível. Isso acaba trazendo uma dificuldade na escolha de quem procura começar um projeto. Com o objetivo de auxiliar nesta escolha, este trabalho faz um estudo comparativo entre algumas plataformas: Apache Singa, Graphlab e H2O. Experimentos são conduzidos utilizando os conjunto de dados MNIST e KDD Cup 1999. Resultados apontam que as plataformas testadas têm suas vantagens: Graphlab é a mais intuitiva, a Apache Singa oferece mais recursos e H2O obteve os melhores resultados de predição.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.