Human Activity Recognition using Binary Motion Image and Deep Learning

Human Activity Recognition using Binary Motion Image and Deep Learning

Abstract— Human Activity Recognition using Binary Motion Image and Deep Learning. Human identification is becoming one of the major worldwide issue now a days. Dental biometrics is the leading biometric technique to identify individuals on the basis of their dental characteristics. Dental Features of persons are naturally unique. They can be used to authenticate humans exactly or almost to the maximum possible similarity. In this paper, we present an efficient workable method to authenticate humans correctly and identify them properly, which is based on dental work information extracted out from dental data. The method we have proposed here comprises of five main processing stages; the initial stage is pre-processing, i.e. initial work on dental data then the Segmentation step, i.e. getting the relevant part of dental data and other processing steps in segmentation. Then Features extraction is performed on segmented images and finally biometric analysis is done which is the most important step for matching. The method is tested on two databases i.e. dental radiographs and colored teeth images and the results are highly encouraging. The data set comprises dental radiographs of 14 persons and colored teeth images of 45 persons. An Equal Error Rate < Final Year Projects 2016 > EER of 85.7% dental radiographs and 88.8% for colored teeth images found on matching the performance of our dental biometric analysis, which shows highly accuracy using our proposed methodology on the data set.