naivebees
Stars
5
Ecosystems:
Python
NaiveBees
This repository provides the winning solution to DrivenData's NaiveBees competition.
Create training file list
python create_training_dataset.py
Run caffe
put bvlc_googlenet.caffemodel in ./
Run:
./PATH_TO_CAFFE/caffe train solver ./solver.prototxt weights ./bvlc_googlenet.caffemodel
snapshots are saved every 1000 iterations to ./snapshots/
Caffe must be run upto 20000 iterations.
Create test predictions
python create_preds.py
##Make final predictions
python create_sub.py