https://github.com/tapishr/kaggle-carvana

Trains a U-net model to perform image segmentation on car images
https://github.com/tapishr/kaggle-carvana

cnn-keras gcp-cloud-functions segmentation u-net

Last synced: 11 days ago
JSON representation

Trains a U-net model to perform image segmentation on car images

• Host: GitHub
• URL: https://github.com/tapishr/kaggle-carvana
• Owner: tapishr
• Created: 2018-01-17T02:00:45.000Z (almost 7 years ago)
• Default Branch: master
• Last Pushed: 2018-01-26T05:23:38.000Z (almost 7 years ago)
• Last Synced: 2024-10-24T22:14:50.733Z (about 2 months ago)
• Topics: cnn-keras, gcp-cloud-functions, segmentation, u-net
• Language: Python
• Size: 1.19 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Kaggle-Carvana

This is a submission for the [Carvana Image segmentation challenge](https://www.kaggle.com/c/carvana-image-masking-challenge/) 

on Kaggle. This solution trains U-nets on Google Cloud Platform to segment car images.

## Instructions

To run the code on a local machine, requirements include -

- Python 2.7

- Tensorflow (with Python 2.7 wrapper)

- OpenCV

- Other libraries that can be installed using `setup.py` script provided with the code

There are 2 sets of code – one for training the models and one for testing. 

Both training and testing code have been designed to run on the Google Cloud Platform (gcp).

To run the code on gcp, run the `start.sh` shell script by typing the following command -

`$ sh start.sh`

Values for arguments in `start.sh` would need to be changed before running it -

- Inside start.sh, the argument `job-dir` must contain the GCS (Google Cloud Storage) URL which will store logs and the model’s weight file. 

- The argument `train-dir` should contain the GCS URL where all the training images and the training mask csv file are stored.

To run the code locally, the script `local.sh` can be executed after giving valid directory paths for `job-dir` and `train-dir` arguments.

### Licence

MIT License

Copyright (c) [2014] [Tapish Rathore]

Permission is hereby granted, free of charge, to any person obtaining a copy

of this software and associated documentation files (the "Software"), to deal

in the Software without restriction, including without limitation the rights

to use, copy, modify, merge, publish, distribute, sublicense, and/or sell

copies of the Software, and to permit persons to whom the Software is

furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all

copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR

IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE

AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER

LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,

OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE

SOFTWARE.