ABSTRACT
According to the Brazilian Institute of Geography and Statistics,
there are about 6.5 million visually impaired in Brazil. A frequent
issue is the insertion of them in hazardous environments signaled
by signs only. Solutions must be produced to improve the life quality
of visually impaired whether at work or in public places. In this
sense, the present paper presents an application of danger sign
recognition aimed at people with visual impairment through a
supervised learning classifier called Haar Cascade. Furthermore,
the paper reports preliminary results regarding effectiveness of
the proposed system applied in a set of images. Overall, for the
majority of the samples, the proposed system achieved a promising
score and this can be a good alternative for studies that aim visually
impaired inclusion in dangerous environments.