Pytorch implementation of COVID-ResNet using GoogleNet (Inception v1). Please see COVID-ResNet readme for details in cloning the datset. The purpose of this model is to provide a slightly lower performance, more efficient model to use on low-powered machines and on the web. To see an online demo, please visit stanleyzheng.ca. This demo is running on less than 200mb of memory, and the entire slug size (including all dependencies) is 239mb.
COVID-ResNet Test:
| Sensivity(%) | |
|---|---|
| Normal | COVID-19 |
| 96.00 | 97.50 |
| Positive Predictive Value (%) | |
|---|---|
| Normal | COVID-19 |
| 95.12 | 97.96 |