BalancedFixMatch

BalancedFixMatch is the model to classify COVID-19 based on Chest X-ray images.

Requirements

• Python 3.7
• PyTorch 1.7.0

Datasets

COVIDx-CXR Dataset (Link)

Type	COVID-19	Pneumonia	Normal	Total
Train	517	5,475	7,966	13,958
Test	100	100	100	300

Performances

Labels	μ	Baseline	FixMatch	FixMatch&Focal loss
150	1	85.25±1.97	85.03±1.08	84.94±1.33
150	2	85.25±1.97	85.81±1.21	86.00±1.46
150	3	85.25±1.97	85.92±1.04	86.25±1.69
300	1	87.06±1.30	87.39±1.41	86.86±1.15
300	2	87.06±1.30	87.25±1.24	87.67±1.28
300	3	87.06±1.30	87.56±1.01	87.78±0.96

How to Train BalancedFixMatch

These are tutorials for training BalancedFixMatch using CXR images:

• Train BalancedFixMatch