chestnut-seaurchin-classifier
Our goal
Our goal is to accurately classify sea urchins and chestnuts.
The ultimate goal is to correctly predict chestnuts in the water
.
Article (Japanese)
https://qiita.com/baibai25/items/93dfd703ccba6e097ba5
Demo
You can easily try to demo.
-
Download model https://drive.google.com/open?id=1Yoq3Nwx72zXTtANvEqp_aNCvoo3wtk_A
-
Run with YOLO
./darknet detector test cfg/foo.data cfg/foo.cfg unikuri_best.weights test_img.jpg
Translation
kuri means chestnut. uni means sea urchin.
Scraping
-k: Search words
-l: Number of images to download
python download.py -k keywords A, keywords B -l 1000
Data augmentation
-i: Input directory
-o: Output directory
python augmentation.py -i input_dir -o output_dir
Building a object detector using YOLO
-
Download
git clone https://github.com/AlexeyAB/darknet.git cd darknet/ make wget https://pjreddie.com/media/files/darknet53.conv.74
-
Move labeled data
-
Define learning settings
cfg/foo.data
cfg/foo.names
cfg/foo.cfg -
Training
./darknet detector train cfg/foo.data cfg/foo.cfg darknet53.conv.74
For more details, refer to the https://github.com/AlexeyAB/darknet?files=1.