Journal of Agricultural Machinery Engineering ISSN:2799-8673(Print) 2799-8819(Online) 농업기계공학

In this paper, a study was conducted on a method of separating the grading part from the cut surface of the beef carcass using the semantic segementation network, which is a kind of artificial intelligence network. In order to apply semantic segmentation of the grading part, the grading part annotated by an expert was used as laerning data. Then, it was divided into 123 training data and 99 test data. Fully convolutional network, Deeplab V3, ResNeSt were applied in the experimental process, and Intersection over Union (IoU) was applied as a performance evaluation index to check which network is suitable for separating the grading part. By applying a labeling algorithm to the resulting image of semantic segmentation network, the part predicted by the semantic segmentation network as a grading part due to a misjudgment is removed. Through this experiment, IoU values were obtained at 0.932, 0.938, and 0.936, respectively.