Skip to content
code from "Automated fynbos identification using iNaturalist and Deep Learning"
Jupyter Notebook Python R
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.gitignore
README.md
imagenet.r
proteas_bottleneck.ipynb
proteas_finetune.ipynb
proteas_predict_single.py

README.md

Automated fynbos identification using iNaturalist and Deep Learning

The code provided in this repo describes the fitting of a convolutional neural network along with this blog post on lentilcurtain.com, my personal blog. Much of the code and workflow is adapted from this post on the keras blog by Francois Chollet.

Before running the model, download VGG16_weight.h5, which are the pretrained model weights from VGG16.

To obtain the training data run imagenet.r
To train the image classifyer run:
proteas_bottleneck.ipynb
then proteas_finetune.ipynb
To make predictions run proteas_predict_single.py

Model build using Keras 2.0 running with tensorflow backend on Python 2.7

You can’t perform that action at this time.