Malaria is caused by parasites and could be transmitted through infected mosquitoes. There are about 200 million cases worldwide, and about 400,000 deaths per year, therefore, malaria does lots of harm to global health.
Although Malaria is a curable disease, inadequate diagnostics make it harder to reduce mortality, as a result, a fast and reliable diagnostic test is a promising and effective way to fight malaria.
To mitigate the problem, we use Singa to implement a machine learning model to help with Malaria diagnosis. The dataset is from Kaggle https://www.kaggle.com/datasets/miracle9to9/files1?resource=download. Please download the dataset before running the scripts.
data
includes the scripts for preprocessing Malaria image datasets.
model
includes the CNN model construction codes by creating a subclass of Module
to wrap the neural network operations of each model.
train_cnn.py
is the training script, which controls the training flow by doing BackPropagation and SGD update.
python train_cnn.py cnn malaria -dir pathToDataset