Multi-label eye fundus image classification system and method and electronic equipment
A fundus image and classification system technology, applied in instruments, biological neural network models, calculations, etc., can solve problems such as poor image feature extraction capabilities, inability to better solve label correlation problems, and low accuracy of classification results
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0067] A multi-label fundus image classification system, characterized in that the system includes:
[0068] an image acquisition device, including a camera and a first communication module;
[0069] An intelligent terminal, including a computer or a mobile phone, wherein the computer or the mobile phone includes an image processing module and a second communication module;
[0070] The cloud server includes an image processing module and a third communication module;
[0071] The described system works in two ways, such as figure 1 The first mode is shown: after the image acquisition device collects the fundus image, the first communication module sends the collected fundus image to the intelligent terminal through the second communication module, and then the image processing module compares the collected fundus image. The fundus image is processed, and the image classification result is displayed on the intelligent terminal;
[0072] Method 2: After the image acquisition...
Embodiment 2
[0109] This embodiment classifies the same set of test sets by using three different splicing methods in four prediction models, wherein the designated image splicing method is Mode 1, the designated feature splicing method is Mode 2, and the designated label splicing method is Mode 3. It can be seen from Table 1 that the four prediction models show a high degree of consistency, and mode 1 has a better prediction effect than mode 2, and mode 2 has a better classification effect than mode 3. Therefore, the present invention selects the image splicing mode of Mode 1 as the fusion mode of the input image.
[0110] Table 1 Comparison of classification effects of different splicing forms under multiple prediction models
[0111]
Embodiment 3
[0113]In this embodiment, the ablation experiments are used to compare the classification effects on the same test set before and after adding the mixed GCN model to the multiple prediction models. It can be seen from Table 2 that all evaluation indicators have been improved after adding the hybrid GCN model.
[0114] Table 2 Comparison of classification effects before and after adding mixed GCN model under multiple prediction models
[0115]
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More