심층신경망을 이용한 실시간 신발갑피 패턴분석 시스템

초록

This study proposes a shoe-upper pattern analysis system to automate shoe manufacturing process. The proposed system consists of detecting and recognizing upper pattern using deep neural network and analyzing the spatial state, such as size and the rotation of the pattern, by the image processing algorithm. In order to evaluate the performance, the test was conducted by an internationally accredited testing laboratory. As a result, we achieved higher performance than the target performance that is 95% of classification rate, 1000msec of processing speed, and less than 3mm of vision precision. The proposed technique will be used in the robot vision system for robot control in the automotive robot system of thseh oes-upper fusing sewing

키워드