Udacity-MLND-Capstone-Project icon indicating copy to clipboard operation
Udacity-MLND-Capstone-Project copied to clipboard

Transfer Learning with VGG16 neural network architecture implemented by Keras on multi-class fish classification problem with data from Nature Conservancy Fishery Monitoring Competition on Kaggle.

Udacity-mlnd-deeplearning-capstone

Transfer Learning with VGG16 neural network architecture on multi-class fish classification problem with data from Nature Conservancy Fishery Monitoring Competition on Kaggle. The final model yilds 1.19 log-loss in the leaderboard with a top-45% ranking currently(so far my best one:D)

The final report with model visualizations and validation plots is here

fish

Dependencies :

  • Python 2.7+/3.5
  • keras pip install keras
  • tensorflow pip install tensorflow

Note that keras and tensorflow have their own dependencies. I recommend using Anaconda for handlinng the packages.

Usage :

  • Download data from Kaggle
  • Use Train-Validation Split Notebook to split train data into proper directories
  • Read the capstone-final report for understanding the approaches taken in solving this problem
  • Notebooks are under the "notebooks" folder and scripts are under the "scripts" folder.
  • All the model diagrams and performance charts are provided.