Skip to main content
. 2020 Sep 18;8:171575–171589. doi: 10.1109/ACCESS.2020.3025010

TABLE 2. Performance Comparison of the Proposed DL-CRC and CNN With Generic and GAN-Based Data Augmentation.

Model Normal Pneumonia COVID-19
CNN-only 97.76% 78.33% 54.55%
CNN & generic 80.48% 23.07% 63.64%
CNN & GAN 88.33% 83.16% 84.84%
Proposed DL-CRC (CNN, GAN, generic) 95.91% 88.52% 93.94%