Unsupervised image retrieval method and system based on label visual joint perception

The invention discloses an unsupervised image retrieval method based on label visual joint perception, which comprises the following steps of: firstly, acquiring an image to be retrieved and a plurality of images in an image database, respectively converting the image to be retrieved and each image into multi-dimensional tensors by using a PIL (Precision Image Library) library, forming a multi-dimensional tensor set by all the multi-dimensional tensors, inputting the multi-dimensional tensor set into a pre-trained retrieval model to obtain a hash code corresponding to the to-be-retrieved image and a hash code corresponding to each image, and calculating the similarity between the hash code corresponding to each image and the hash code corresponding to the to-be-retrieved image to obtain a retrieval result; wherein the retrieval model comprises a ResNet-101 network, a label co-occurrence embedding generation network, a visual similarity embedding generation network, a dual OGSP network, a self-supervision netw