- python==3.6
- torch==0.4.1
- torchvision==0.2.1
other dependencies in requierements.txt
Forked and Adapted from: https://github.com/earhian/Humpback-Whale-Identification-1st- https://www.kaggle.com/c/humpback-whale-identification/discussion/82366
trainset -> ./WC_input/train
testset -> ./WC_input/test
cd input
Download training masks
https://storage.googleapis.com/kaggle-forum-message-attachments/459392/11072/masks.zip
unzip mask.zip
download humpback whale kaggle data, then put them into input
https://www.kaggle.com/c/whale-categorization-playground/data
download sperm whale data, flatten the data in single directories train, test, and data (all) then put them into WC_input https://gdsc.ce.capgemini.com/resources/
line 345 in train.py
step 1.
freeze = False
model_name = 'seresnext101'
min_num_class = 10
checkPoint_start = 0
lr = 3e-4
#until train map5 >= 0.98
step 2.
freeze = True
model_name = 'seresnext101'
min_num_class = 0
checkPoint_start = 0
lr = 3e-4
step 2.
freeze = True
model_name = 'seresnext101'
min_num_class = 0
checkPoint_start = 0
lr = 3e-4
test_train=True
line 130 in test.py
checkPoint_start = best checkPoint of step 2