Fig. 5
From: Acquire continuous and precise score for fundus image quality assessment: FTHNet and FQS dataset
The confusion matrices of internal and external dataset test. We applied our pretrained model to (a) the FQS dataset and (b) EyeQ44 dataset. The image quality level is decided by its quality score prediction following the same criteria as in Section.1.1.2. The FTHNet achieves great performance in quality level classification on internal dataset validation. The external test also indicates that FTHNet works effectively on the external dataset. The results of classification show great precision and generalizability of our FTHNet.
