Open world target detection method, computer device and storage medium

Abstract

This invention provides an open-world object detection method, computer device, and storage medium. The method includes: acquiring image data to be identified and corresponding target text labels; determining the target region of interest (ROI) of the image data to be identified, and extracting features from the ROI using a pre-trained image model to obtain a target region graph vector; determining the category information corresponding to the target text labels, and extracting features from the category information using a pre-trained text model to obtain target word vectors; and analyzing the target region graph vectors and target word vectors using a target neural network model to obtain the identification result of the image data to be identified, wherein the target neural network is an MLP (Multilayer Perceptron) network. This application aims to achieve image recognition based on image data and its corresponding text labels, thereby improving the accuracy and efficiency of the recognition results. Especially for the recognition of medical images, it can reduce the difficulty of recognition.

Images