Abstract: A fast and effective product classification method based on deep learning technology, MdmNet, is proposed, which embeds human's motivation thinking mode into machine learning to form a new product design method. By using the consumer information reasoning method and the weighted fusion module, an experiment is carried out on the benchmark data set Cars. According to the product characteristics of the producer, a classification of motive requirements based on multi-neural network feature fusion is proposed. This idea is to use MDM and multi-neural network fusion to increase the feature extraction of image database data. This method can significantly improve the performance of new product classification, compared with the comparison model, its performance is reflected in the accuracy and other indicators. On the benchmark Dataset Cars, when set to 0.1 and 0.9, MdmNet outputs Top-1, Top-5, and Top-10 correct rates are higher than those of traditional algorithms from the designer's point of view. MdmNet is suitable for such system objects with small amount of data and short time.

Key words: motivation design model, deep learning, product classification, new product development, target consumers