Combinational Class Activation Maps for Weakly Supervised Object Localization (WACV 2020)
Combinational Class Activation Maps for Weakly Supervised Object Localization
Environments
python == 3.6
pytorch == 0.4.0
opencv == 3.4.2
Dataset
CUB-200-2011
PATH: ./CUB_200_2011
Pre-trained model
Training and Test for your own model
Change codes for your model
Training: train_cub.py
- model
- dataloader
Evaluating: IoU_check_cub_5.py
- load model
- choose function for CCAM
Get the paper results
For getting bounding boxes, we should use a threshold value.
I fix the threshold value for quadratic function.
You can change threshold values for the best results of each combinational function.