This repo contains a brain4k pipeline for a kitten-or-bear classifier. It is intended to serve as an example of how to train a classifier on imagenet features. The features are the activity of the penultimate layer of a convolutional neural network trained on the imagenet database.
to execute the pipeline within a docker container:
sudo docker run -ti tleyden5iwx/caffe-cpu-master /bin/bash
pip install git+https://github.com/shuggiefisher/brain4k.git
git clone https://github.com/shuggiefisher/kitten-or-bear-classifier.git local-path-to-this-repo
brain4k local-path-to-this-repo
- brain4k
- caffe, including python wrappers
- scikit-learn
- matplotlib
Installing caffe is quite involved, so you may find it easier to use a caffe docker image
Confusion matrix:
[[11 4] [12 10]]