Depth semi-supervised image classification method based on discriminant feature learning and entropy

The invention discloses a deep semi-supervised image classification method based on discriminant feature learning and entropy. The method comprises the following steps: dividing a training sample set into a labeled training sample set and an unlabeled training sample set; building a deep convolutional neural network model; constructing a discriminant feature learning objective function; constructing an entropy objective function; constructing a total objective function; setting the value of the training number of the current round as 0; training the constructed network model until the model converges; based on the current network model, calculating the probability that each unlabeled sample belongs to different classes; updating a labeled training sample set and an unlabeled training sample set; adding 1 to the value of the training number of the current round; repeating the steps 7)-10) until the training number of the current round reaches the preset maximum training number of the round; and inputting a to-