Acne identification method based on improved YOLOv4 algorithm

• Main Authors: DONG LEI, MA SHULONG, HUANG YANG
• Format: Patent
• Language: Chinese; English
• Published: 24.01.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention particularly relates to an acne identification method based on an improved YOLOv4 algorithm. According to the acne identification method based on the improved YOLOv4 algorithm, a data set is divided into a training set and a test set in a user-defined mode, data of the training set is enriched through a data enhancement method, and the robustness of a training model is enhanced; processing the enhanced training set data by using a K-means clustering method to obtain an anchor point anchors size suitable for acne detection; and performing iterative training on the YOLOv4 algorithm model by using the enhanced training set, and testing the test set by using the trained YOLOv4 algorithm model. According to the acne identification method based on the improved YOLOv4 algorithm, by improving a YOLOv4 algorithm model, the method can better accord with detection and identification of small targets, the acnes are classified according to the severity of the acnes, and then reference is provided for a user.