Across Domains models on garbage classification
Hanpeng Li · 2021 International Conference on Computer Information Science and Artificial Intelligence (CISAI) · 2021
The existence of deep learning has solved many problems for human beings. In order to solve the limitations of the deep learning model when converting between different domains, a separation neural network (MIDNet) based on mutual information is proposed to realize the transfer of knowledge to invisible classification features in the target domain. MIDNet was developed as a semi-supervised learning model, which requires almost no new data labeling when faced with a new domain. It can play an important role when faced with tasks that require a large amount of data annotation, such as garbage classification. In view of the current intelligent development of garbage classification has reached a climax, applying it to the field of garbage classification can improve the efficiency and accuracy of classification. At the same time, the cost of manually labeling data can be reduced. This article will explain the specific application of the model in garbage classification, and discuss the changes it can bring and the future development direction.