https://github.com/lromul/kaggle-planet-amazon
Solution for Kaggle competition "Planet: Understanding the Amazon from Space"
https://github.com/lromul/kaggle-planet-amazon
Last synced: 6 months ago
JSON representation
Solution for Kaggle competition "Planet: Understanding the Amazon from Space"
- Host: GitHub
- URL: https://github.com/lromul/kaggle-planet-amazon
- Owner: lRomul
- Created: 2017-08-31T09:15:30.000Z (about 8 years ago)
- Default Branch: master
- Last Pushed: 2017-09-01T13:44:28.000Z (about 8 years ago)
- Last Synced: 2025-03-27T07:48:57.442Z (6 months ago)
- Language: Python
- Size: 126 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
Planet: Understanding the Amazon from Space
===========================================
Use satellite data to track the human footprint in the Amazon rainforest.
This is my part of our team's solution for the Kaggle challange of
[Understanding the Amazon from Space](https://www.kaggle.com/c/planet-understanding-the-amazon-from-space).
Our team ods.ai [finished 7th](https://www.kaggle.com/c/planet-understanding-the-amazon-from-space/leaderboard/private).
## Requirements
* Linux
* Nvidia drivers, CUDA 8
* Docker, nvidia-docker
## How to Use?
Put data to ``data``::
data
├── train-jpg
├── test
├── 10_folds.npy
├── train_v2.csv
└── sample_submission_v2.csv
`test` contains images from `test-jpg` and `test-jpg-additional`.
1. Go to folder `docker` and build image
```
cd docker
./build.sh
```
2. Run container with nvidia-docker
```
./run.sh
```
3. Go to folder `src/kfold_train` and start train models
```
cd src/kfold_train/
python densenet121_001.py
python vgg11_001.py
...
python vgg19_001.py
```