Fastener Classification Using One-Shot Learning with Siamese Convolution Networks

Deep Learning has been widely used in image-based applications such as object classification, detection, and recognition recent years. Classifying highly similar objects is a very difficult problem. It to classify datasets this situation where similarity between classes differences are high. In study, Siamese Convolution Neural Network, which measurement-based network, practiced 6 types of screws, 5 nuts, 7 bolts that each other. addition, neural network formed with the One-Shot technique trained. Thanks OSL technique, there no need use large data sets. Also, amounts from class. Adding new class be classified also made easier by technique. The performance results proposed method manifested detail article.