Research And Implementation Of E- Commerce Product Image Recognition System Based On Deep Learning

With the rapid development of the Internet, e-commerce is rapidly rising, providing convenient services in e-commerce has become a hot research topic. At the same time, the image recognition technology based on deep learning has become a new research and application hotspot in the field of artificial intelligence with the characteristics of high accuracy and scalability. Combining the image recognition technology based on deep learning and the actual e-commerce products to analyze application requirements of the current e-commerce product market, the paper will take a large number of pictures of e-commerce products as data sets and use the Convolutional neural network algorithm based on deep learning of image mining, finally realizes the online image recognition tasks. For example, the field of medicine is used as a testing area in this study. First of all, a large number of medicine images are cleaned, extracted and expanded, and the original data sets of medicine images are constructed.Secondly, the original data set of the drug image is extracted by using the Convolutional neural network algorithm based on deep learning and construct the training model of deep learning. Finally, the training model is used to extract the feature and calculate the probability, and then the corresponding medicine name is identified and the corresponding medicine is retrieved to show. The system will provide great convenience to the elderly or to a group of customers who are unfamiliar with a medicine and will be conducive to the promotion of commercial products and enhance customer satisfaction.In order to realize the image recognition and retrieve corresponding products and show it, this paper studies and realizes the system from two aspects:First of all, the convolutional neural network in deep learning as the prototype, using the deep learning platform Caffe to carry out a lot of model training on the medicine data set of this article, the model combination is applied to three kinds of pattern recognition: Do not distinguish between large class recognition, distinguish between large class recognition and the group BLOCK identification. The three patterns were adapted to classes of data sets from less and multiple case, innovative to meet the majority of medicine identification tasks and has better application and accuracy.Secondly, based on the above three patterns, the Web system, which is based on the construction of the image data set and the text data set, and the SSH architecture, is used to build the online image recognition and retrieval system based on deep learning. Through the online simulation system, intuitive show the operability and the development of e-commerce to help promote the role which combines the image recognition technology based on deep learning with e-commerce products.Of course, the image recognition technology based on deep learning is currently in the initial stage of development, and This paper also summarizes the achievements and shortcomings of the work done. In addition, the future work of this paper is further thinking and prospect.