Product image optimization method and system based on intelligent e-commerce platform

• Main Authors: JI GUOHONG, ZHAO CHANGXIU, ZENG BIN
• Format: Patent
• Language: Chinese; English
• Published: 13.10.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a product image optimization method based on an intelligent e-commerce platform. The method comprises the steps of obtaining a product image data set; initializing a ResNet model, and performing training based on the product image data set to complete classification of the product images; a model distillation method is applied, knowledge of the trained ResNet model is transferred to an optimized student model, and the optimized student model is distributed in a plurality of edge nodes; periodically collecting updated product image data, and adding the updated product image data into the product image data set; training the updated product image data by using an incremental learning method, and regularly updating the optimized student model; and selecting a part of unlabeled samples for labeling according to the confidence coefficient of the optimized student model by using an active learning method, adding the labeled samples into the product image data set, and further optimizing the