Skip to content

gsdndeer/Anemic-Nail-Grading-and-Prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Anemic-Nail-Grading-and-Prediction

Now, hemoglobin values are usually obtained by drawing blood. But it's time-comsuing. I evaluated hemoglobin values directly from nail images in this repository.

I. Grading

Grade
x >= 8 	        => 1
11 >= x > 8    	=> 2
x > 11 	        => 3

I used classification to grade nail images.

Model number Backbone Image size Optimizer Learning rate Criterion Vadlidation accuracy
1 EffcientNet-b7 512x386 Adam CEL 0.001 0.89
2 EffcientNet-b7 512x386 Ranger CEL 0.001 0.87
3 EffcientNet-b7 512x386 SGD CEL 0.001 0.83

II. Prediction

Model number Backbone Image size Optimizer Learning rate Criterion Vadlidation accuracy
1 EffcientNet-b2 512x386 Adam 0.001 MSE 0.90
2 EffcientNet-b7 512x386 Adam MSE 0.001 0.92
3 EffcientNet-b7 512x386 Ranger MSE 0.001 0.91

Acknowledgement

  1. EfficientNet PyTorch

Releases

No releases published

Packages

No packages published