bryant-pham / cosc6373-cv-ibtl

Requirements:

• Python 3.6.5

To install:

pip install -r requirements.txt

Dogs and cats dataset: https://www.kaggle.com/c/dogs-vs-cats/data

Monkey species dataset: https://www.kaggle.com/slothkong/10-monkey-species

Special note on Monkey Species dataset: Contains 10 different species of monkey. In order to specify a specifc class for binary classification, training and validation sets should be made such that one species is one class and all other species are the 2nd class

Experiment scripts:

• dogscats_base.py
• dogscats_ibtl.py
• monkey_base.py
• monkey_ibtl.py

How to run Dogs vs Cats experiment:

1. Download Dogs vs Cats data: https://www.kaggle.com/c/dogs-vs-cats/data
2. Extract dog training images into data/dogscats/train/dogs
3. Extract cat training images into data/dogscats/train/cats
4. Extract dog validation images into data/dogscats/validation/dogs
5. Extract cat validation images into data/dogscats/validation/cat
6. Run:

python dogscats_base.py
python dogscats_ibtl.py

How to run Monkey Species experiment:

1. Download Monkey Species data: https://www.kaggle.com/slothkong/10-monkey-species
2. Extract monkey training image folders into data/monkey/train
3. Extract monkey validation images into data/monkey/validation
4. Run:

python monkey_base.py
python monkey_ibtl.py

How to predict monkey species:

1. Download Monkey Species data: https://www.kaggle.com/slothkong/10-monkey-species
2. Run:

python predict_monkey.py <path_to_monkey_image>