Cross-domain image classification method based on pseudo label domain adaptation

Abstract

The invention discloses a cross-domain image classification method based on pseudo label domain adaptation, and relates to image processing. The method comprises the following steps: A, pre-training a depth model; b, generating a target domain image pseudo label; and C, training and optimizing the depth model. The method is simple and effective, and a better effect can be obtained on a plurality of natural image data sets. The influence of false labels is avoided; the pseudo label samples are added into the training set to further train the network, so that the discrimination of the model in the target domain can be improved, and the generalization performance of the model in the target domain is improved. The target domain data without labels are fully utilized by alternately making false labels on the target domain and training the network by adopting false label samples, so that the recognition performance of the model on the target domain is effectively improved. The invention improves the generalization performance when a deep learning model is used for natural image recognition, improves the accuracy and recall rate during cross-database testing, is high in practicability and portability, and can meet the requirements of weak supervision learning under the conditions of large domain difference, unbalanced categories and the like.

Description

technical field [0001] The present invention relates to image processing, in particular to a cross-domain image classification method based on pseudo-label domain adaptation that can improve the generalization performance when using a deep learning model for natural image recognition, and improve the accuracy and recall of cross-database testing. Background technique [0002] Deep convolutional networks combine feature extractors and classifiers to learn discriminative features through end-to-end training. In recent years, with the rapid development of computer computing power and machine learning theory, the recognition accuracy of natural images has increased rapidly. In large-scale data sets (such as ImageNet data sets) (Deng J, Dong W, Socher R, et al. Imagenet: A large-scale hierarchical image database[C] / / 2009IEEE conference on computer vision and pattern recognition.Ieee,2009:248-255) the recognition performance of the deep convolutional network far exceeds the tradit...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve practical problems such as difficulty in obtaining data and expensive labeling data in the prior art, and provide a method that can improve the generalization performance of natural image recognition using deep learning models, and improve the accuracy and recall of cross-database testing. A Cross-Domain Image Classification Method Based on Pseudo-Label Domain Adaptation

Images