Overview / Usage

In 2018, Rajaraman et al. published a paper entitled Pre-trained convolutional neural networks as feature extractors toward improved parasite detection in thin blood smear images. This was a work to showcase that it is possible to use deep learning to detect malaria by detecting the parasite in thin blood smear images,

In their work Rajaraman et al. utilized six pre-trained Convolutional Neural Networks, including:

• AlexNet
• VGG-16
• ResNet-50
• Xception
• DenseNet-121
• A customized model they created

Feature extraction and subsequent training took a little over 24 hours and obtained an impressive 95.9% accuracy.

The problem here is the number of models being utilized — it’s inefficient.

Methodology / Approach

• Carefully constructed data augmentation pipelines
• Used a pre-trained ResNet34 architecture
• Fine-tuned the last layers of the model
• Incorporated modern deep learning practices like discriminative learning rates, mixed precision policy and so on
• Implemented classification reports to present the results in a decent way

Technologies Used

• Python
• fastai, numpy, sklearn
• Google Colab

Repository

https://github.com/sayakpaul/Malaria-Detection-with-Deep-Learning