Menu
As a folk traditional handicraft with a long history in Guangxi, mango weaving craft is facing the dilemma of shrinkage and decline. Starting from the application of AI technology in the protection and innovation of mang weaving technology, the article analyzes the text of collected mang weaving product reviews with the help of the perceived value theory and LDA topic model, and mines the consumer preference of mang weaving in order to better improve the production and manufacturing of mang weaving. Meanwhile, a lightweight generative adversarial network based on non-local attention mechanism is proposed to realize the construction of text-to-multi-objective image generation model, which is used in the innovative design of mango knitting. Consumers are more satisfied with the overall experience of mango knitting, and positive emotions account for 82.6%. The review texts were categorized into five themes with probability values of 0.181 to 0.227, and improvements should be made to the mango knitting product in terms of emotion, quality and price. The text-generated image model is able to output the mango knitting design images well, and the average values of IS and FID are 21.85% and 16.46% better than the comparison model, and the quality of the output images is better than the other models in different test sets. With the help of AI technology to improve the process of mango weaving, to enhance the design creativity of product quality, and to broaden the channels for the inheritance and development of mango weaving skills.