Eye fundus image recognition method and device, electronic equipment and storage medium

[0035] figure 1 It is a flowchart of a fundus image recognition method in Embodiment 1 of the present invention. The embodiment of the present invention is applicable to the situation of identifying the state of the lesion on the fundus image. The method is implemented by the fundus image recognition device. The device is implemented by software and / or hardware, and is specifically configured in an electronic device. The electronic device may have certain A mobile terminal or a fixed terminal with data processing capability, and also a server.

[0036] like figure 1 A fundus image recognition method shown includes:

[0037] S110. Acquire a fundus image to be identified, and perform grid block processing on the fundus image to be identified to form a plurality of block grids to be identified.

[0038] Among them, the fundus is composed of the retina, fundus blood vessels, optic nerve head, optic nerve fibers, macula on the retina, and choroid behind the retina. Fundus images

[0054] figure 2 It is a flowchart of a fundus image recognition method in Embodiment 2 of the present invention. The embodiments of the present invention are optimized and improved on the basis of the technical solutions of the foregoing embodiments.

[0055] Further, before the operation "input each block grid of the fundus image to be identified into the pre-trained neural network model to determine the lesion state of the block grid to be identified", add " Obtain the block grid of lesions and the block grid of ordinary people, wherein, the block grid of the focus is marked in the fundus image of the lesion and include the block grid of lesions; the block grid of the focus and the block grid of ordinary people Grid input in the neural network model for training", to improve the training mechanism of the neural network model.

[0056] like figure 2 A fundus image recognition method shown includes:

[0057] S211. Obtain a lesion segmented grid and an ordinary person's segm

[0078] On the basis of the technical solutions of the foregoing embodiments, the embodiment of the present invention optimizes and improves the fundus image recognition method.

[0079] Figure 3A It is a schematic structural diagram of a fundus image recognition model in Embodiment 3 of the present invention. The fundus image recognition model includes a preprocessing module 310 , a model training module 320 and a postprocessing module 330 .

[0080] The preprocessing module 310 is configured to perform preprocessing operations on the acquired reference fundus images. Wherein the reference fundus image can be a fundus image to be recognized, or a training fundus image during model training; the training fundus image includes a focus fundus image for performing lesion block grid extraction, and the user performs ordinary person block network Fundus images of normal people obtained by grid.

[0081] Specifically, the preprocessing operation on the acquired reference fundus imag