Small target detection method based on image two-dimensional entropy segmentation and knowledge distillation

Abstract

The application provides a kind of small target detection method based on image two-dimensional entropy segmentation and knowledge distillation, belongs to image small target detection technical field;In order to solve the technical problems of insufficient small target feature resolution, lack of significant features and insufficient model generalization ability in the existing small target detection technology in complex traffic scenes, the technical scheme adopted is: a small target detection model is constructed in the automatic driving environment of complex traffic scenes;Based on the candidate region selection method of image two-dimensional entropy and density estimation, an image segmentation selection strategy is designed;The pre-trained visual language model is used as a teacher model for knowledge distillation, and the knowledge is transferred to a lightweight student model to build a target detection head;A two-stage target detection model is constructed, and potential small targets in complex traffic scenes are detected based on the target detection model;The application is applied to the small target detection of automatic driving.

Images