Object recognition for control panels on machine tools with HOG and Bag of Keypoints

Abstract

Currently, the components in the control panel for machine tools, electrical wiring connections are called harnesses performed manually. Therefore, it is required by a machine to automate its work. The purpose of this study is performing by image recognition of these two things. First, detecting the pre-installed screw with components in order to wire harness automatically. Second, scanning the connected harness after installing the harness. In this study, we use a technique called generic object recognition which learns and classifies the image feature by means of machine learning. We use HOG (Histograms of Oriented Gradients) and Bag of Keypoints as a method of calculation for the feature, AdaBoost and SVM (Support Vector Machine) as a method of machine learning. In this paper, we show the detection rate of screws and harnesses using the method described above.