Predict horse's distance aptitude (sprint / mile / middle / long) from their padock photo images.

• 0.0 - 0.2: sprint
• 0.2 - 0.4: mile
• 0.4 - 0.6: middle
• 0.6 - 1.0: long

• Python3
• chainer
• numpy
• PIL
• tqdm

1. Prepare padock photo images and store them into data/original/ directory.

   or you can download my sample dataset.

   $ cd data
   $ ./download_sample.sh
   $ unzip keibado_padock_images.zip
   $ find keibado_padock_images -type f | xargs -J% cp % original/
   

2. Preprocess images

   $ ./preprocess_all.sh
   

3. Generate a mean image

   $ cd ../
   $ ./compute_mean.sh 
   

4. Train dataset

   $ train.sh
   

Output loss for test images.

$ eval.sh result/your_model_output