Target detection network training method and device, equipment and storage medium

Abstract

The embodiment of the invention provides a target detection network training method and device, equipment and a storage medium, and relates to the technical field of image processing. The method comprises the following steps: acquiring a reference intermediate feature and a corresponding initial intermediate feature, at least inputting the initial intermediate feature into a corresponding feature fusion network for feature fusion to obtain a fused feature, and calculating an intermediate loss value corresponding to a corresponding initial residual block according to the fused feature and the reference intermediate feature, the input data of the last feature fusion network is the corresponding initial intermediate feature, and the input data of other feature fusion networks are the fusion features corresponding to the initial intermediate feature and the next feature fusion network. And updating the initial target detection network at least based on the intermediate loss value. A feature fusion network is introduced in a teacher-student consistency distillation learning process, and aggregation operation is performed on multi-level features obtained in a self-training process, so that cross-level interaction of the features is realized, teacher-student consistency regularization is achieved, and the detection accuracy of a target detection network is improved.

Images