AiScope

http://aiscope.net/

---

Detecting malaria parasites in thick blood samples using retinanet object detection architecture https://arxiv.org/abs/1708.02002

Setup

This package requires

• https://github.com/i008/detdata
• https://github.com/i008/keras-retinanet

Install in develop mode:

python setup.py develop

Inference

Check the ./notebooks directory for examples how to predict and train the detector

Example (inference)

Augmentation

Augmentation is done in detdata using the amazing imgaug package (https://github.com/aleju/imgaug)

Run the demo (docker required)

To test on other images - place them in the testimages folder.

git clone https://github.com/i008/aiscope.git && cd aiscope

# run the container
sudo docker run -p 7777:7777 \
-v $PWD/testimages:/opt/aiscope/images \
-t i008/aiscope:latest jupyter notebook \
--no-browser \
--port 7777 \
--allow-root \
--ip=0.0.0.0 \
--NotebookApp.token=''

#go to localhost:7777
#open, run, play with demo.ipynb