Network training method and device, electronic equipment and storage medium

The invention provides a network training method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a training sample set of a target network for carrying out the pixel-level classification of an image; determining a first loss according to the fact that a sample similarity between indication unmarked pixel samples corresponding to any two unmarked pixel samples in the training sample set is a first similarity coefficient and a first mark difference between prediction category marks output by the target network for any two unmarked pixel samples; determining a second loss according to a second similarity coefficient corresponding to any two marked pixel samples in the training sample set and indicating intra-class sample similarity or inter-class sample similarity between the marked pixel samples, and a second mark difference between prediction class marks output by the target network for any two marked pixel samples; and determining a target loss based on the