Integrated collaborative training method and device for zero sample classification and terminal equipment

Abstract

The invention relates to an integrated collaborative training method and device for zero sample classification and terminal equipment, and the method comprises the steps: dividing an obtained data setinto a training set and a test set, calling the training set and the test set as a visible class and an invisible class, training attribute prediction networks of different structures, selecting twonetworks as a main network and an auxiliary network, calculating attribute mapping parameters, synthesizing virtual features of invisible classes according to the attribute mapping parameters, combining the virtual features with a plurality of classifiers to complete training of the classifiers, extracting the features of the invisible classes by using a main network and an auxiliary network, predicting the features of the invisible classes by using the classifiers, and endowing the invisible classes meeting conditions with pseudo tags according to a classifier voting mechanism; adding invisible classes endowed with pseudo tags into a training set to train the attribute prediction network again so that prediction precision of a network model is improved when different ZSL embedding methodscan be used for training to select a main network and an auxiliary network. The invention easily expands to other zero sample learning methods and performance is improved.

Description

technical field [0001] The invention relates to an integrated collaborative training method, device and terminal equipment for zero-sample classification. Background technique [0002] Thanks to the effectiveness of deep learning in image recognition problems, supervised image recognition methods have achieved amazing results in many fields, but they often require a considerable number of labeled samples to train a good enough network recognition Model, and the model trained by using known samples can only recognize the object classes contained in the training set, and lacks the ability to identify object classes not included in the training set. However, in real life, some categories of image data are scarce and the number of image categories that need to be recognized is increasing. At the same time, the cost of retraining the model is relatively high every time a different category of data is added. The field of image recognition should not rely entirely on this need. sa...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide an integrated collaborative training method, device and terminal equipment for zero-sample classification to solve the problem of poor prediction accuracy of the network model obtained by the existing training method

Images