TEST TASK IN DATA SCIENCE

During working in the company you will be faced with different computer vision tasks. So it is useful to have skills in image segmentation. After completing this test task you will be able to implement similar algorithms in commercial projects.

​

​

Current task requires solving one of the problems from the Kaggle platform: Airbus Ship Detection Challenge.

​
 

You can find the dataset here: dataset for test task.

​
 

The goal of the test project is to build a semantic segmentation model. Prefered tools and notes: tf.keras, Unet architecture for neural network, dice score, python.

​

 

Results of your work should contain next:

● link to GitHub repository with all source codes;
● code for model training and model inference should be separated into
different .py files;
● readme.md file with complete description of solution;
● requirements.txt with required python modules;
● jupyter notebook with exploratory data analysis of the dataset;
● any other things used during the working with task;
 

● Source code should be well readable and commented;
● Project has to be easy to deploy for testing.

​

Good luck and happy working!