Pseudo tag generation method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and provides a pseudo tag generation method and device, electronic equipment and a storage medium. The method is applied to an unmanned vehicle, namely unmanned equipment or automatic driving equipment, and comprises the following steps: training a network model by using labeled data in a point cloud data set; executing a reasoning step to perform reasoning on the unlabeled data in the point cloud data set by using the network model obtained by training to obtain the confidence of the unlabeled data and uncertainty estimation of the confidence; calculating a new confidence coefficient based on the confidence coefficient and uncertainty estimation of the confidence coefficient; based on the new confidence coefficient, performing pseudo-label processing on the non-label data to generate pseudo-label data of the non-label data; executing a second training step to train the network model by using the pseudo label data and the labeled data; and alternately executing the reasoning step and the second training step until the training ending condition of the network model is met. The generalization performance of the network model is improved.

Description

technical field [0001] The present disclosure relates to the technical field of artificial intelligence, and in particular to a pseudo-label generation method, device, electronic equipment, and computer-readable storage medium. Background technique [0002] Pseudo-labeling is a common technique in supervised machine learning. It refers to using labeled data to train a model, and then using the trained model to predict labels for unlabeled data. The predicted labels are pseudo-labels. Generating pseudo labels can add some unlabeled data to the labeled data to jointly train the model to improve the quality of the model. [0003] At present, most of the supervised machine learning algorithms are trained with the maximum posterior, which often produces a point estimate rather than an uncertain value. Specifically, the probability vector after the Softmax layer (an activation function in a neural network) can be used to explain model confidence. But in fact, the model will stil...

AI Technical Summary

Problems solved by technology

[0004] In view of this, the embodiments of the present disclosure provide a pseudo-label generation method, device, electronic device and computer-readable storage medium to solve the problem that a large number of unlabeled samples are labeled with wrong labels and used for training in the prior art, resulting in There are a large number of noise samples in the training set, which seriously affects the performance of the model

Images