http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
주조품 분류 정확도 향상을 위한 적대적 생성 네트워크(GAN) 이미지 데이터 확장
이준협 ( Junhyup Lee ),박기홍 ( Keehong Park ),은준엽 ( Joonyup Eun ) 한국로지스틱스학회 2021 로지스틱스연구 Vol.29 No.4
Casting is a method of making metal through molds by changing solid metal into liquid state, which has the advantage of being able to yield a large number of products with complex shapes. Owing to the advantage, it is widely used to manufacture jewelry, artwork, surgical implants, and impellers in automobiles and ships. However, low quality products can be produced due to pinholes, sand blows, shrinkage cavities, and cracks that are well-known issues in casting. Especially using a defective impeller, a rotating element of a centrifugal pump that accelerates fluid outside from the center and transfers the power of fluid kinetic energy, causes a significant damage to its pump and/or workers nearby due to its high pressure. Therefore, foundries endeavor to catch any defectives before sending them out to purchasers. However, foundries are usually small or medium-sized enterprises. It is difficult for them to hire additional experienced workers to catch more defectives or install photographing and imaging-storing devices to keep track of a large amount of product images for analyses. The foundries usually have a few inspectors to catch defective products and, due to a shortage of manpower and human inaccuracy, defective products are often classified as non-defective products. This study shows that a combination of classic augmentation and self-attention generative adversarial network improves the accuracy of classifying non-defective and defective impellers by augmenting a limited amount of image data that can be even manually photographed. Combining classic augmentation and self-attention generative adversarial network outperforms the sole use of classic augmentation in generating quality images for convolutional neural network.