https://github.com/wdhorton/protein-atlas-fastai

Code for training a Resnet model for the Human Protein Atlas Image Classification competition
https://github.com/wdhorton/protein-atlas-fastai

Last synced: 3 months ago
JSON representation

Code for training a Resnet model for the Human Protein Atlas Image Classification competition

• Host: GitHub
• URL: https://github.com/wdhorton/protein-atlas-fastai
• Owner: wdhorton
• Created: 2018-11-09T13:58:48.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2018-11-20T03:15:19.000Z (almost 6 years ago)
• Last Synced: 2024-07-03T04:04:15.417Z (4 months ago)
• Language: Jupyter Notebook
• Homepage:
• Size: 1.48 MB
• Stars: 50
• Watchers: 4
• Forks: 12
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-fastai - Protein Atlas fastai
• awesome-fastai - Protein Atlas fastai

README

        
# Human Protein Atlas Image Classification Starter Pack

This is the code you need to train a resnet model and submit to the Human Protein Atlas competition. 

It uses the newest version (v1) of the [fastai library](https://github.com/fastai/fastai).

To download the data, I used the [Official Kaggle API package](https://github.com/Kaggle/kaggle-api), which you can install with pip. Once installed, you can run `kaggle competitions download -c human-protein-atlas-image-classification` to get the data (just make sure to update the `path` variable in the `resnet50_basic.ipynb` notebook to point to the data on your machine).

Update 11/17 -- the resnet50_basic notebook doesn't work with fastai version 1.0.25 and above, so I made another notebook to work with the new `data_block` API. In this version, I also made changes to use the `create_cnn` function. You can find it at resnet50_basic_datablocks.ipynb.