Image feature identification method and apparatus, storage medium, and electronic apparatus

Abstract

The invention discloses an image feature identification method and apparatus, a storage medium, and an electronic apparatus. The method comprises the steps of obtaining an identification request, wherein the identification request is used for requesting to identify image features in a target image; identifying the image features in the target image through a first neural network model, wherein thefirst neural network model is obtained after parameters in a second neural network model are trained through a first training set and a second training set, image features of training images in the first training set are marked, and image features of training images in the second training set are not marked; and returning a first identification result of the first neural network model, wherein the first identification result is at least used for indicating the image features (such as lesion features) identified in the target image. The technical problem of relatively low screening efficiencyof diabetic retinopathy in related technologies is solved.

Description

technical field [0001] The present invention relates to the field of the Internet, in particular, to an image feature recognition method and device, a storage medium, and an electronic device. Background technique [0002] Diabetic retinopathy (DR) is the most important manifestation of diabetic microangiopathy, a fundus lesion with specific changes, and one of the serious complications of diabetes. Including non-proliferative diabetic retinopathy (NPDR) (or simple or background type) and proliferative diabetic retinopathy (PDR) and other types. According to the World Health Organization (WHO), by 2030, the number of DR patients in the world will increase to 366 million, and it has become one of the four major blindness diseases. The prevention and treatment of DR will become a serious worldwide problem. [0003] Studies have shown that early diagnosis and treatment of DR patients can effectively prevent visual loss and blindness, and the key to prevention is through fundus...

AI Technical Summary

Problems solved by technology

However, more than 50% of diabetic patients in the world have not received any form of eye examination, and the screening of diabetic retinopathy based on fundus images basically relies on the naked eye observation of ophthalmologists.

[0004] However, when faced with large-scale screening, the amount of data that needs to be analyzed and processed by doctors is very large, and the manual interpretation method is time-consuming and laborious, and manual screening cannot be implemented

Moreover, manual screening is highly subjective, data analysis is complex and difficult to quantify, and it is difficult to achieve quantitative follow-up

[0005] Aiming at the technical problem of low screening efficiency for diabetic retinopathy in related technologies, no effective solution has been proposed so far

Images