Skip to content

Referring Expression Object Segmentation with Caption-Aware Consistency, BMVC 2019

License

Notifications You must be signed in to change notification settings

wenz116/lang2seg

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

lang2seg

Training

  1. Train the baseline segmentation model:
./experiments/scripts/train_baseline.sh GPUID refcoco unc

The Mask-RCNN model is in pyutils/mask-faster-rcnn/lib/nets/resnet_v1.py and pyutils/mask-faster-rcnn/lib/nets/network.py.

  1. Train the whole model with caption loss:
./experiments/scripts/train_cycle.sh GPUID refcoco unc att2in2 CAPTION_LOSS_WEIGHT

The whole model is in pyutils/mask-faster-rcnn/lib/nets/resnet_v1_cycle.py and pyutils/mask-faster-rcnn/lib/nets/network_cycle.py.

Losses are calculated in _add_losses().

Evaluation

./experiments/scripts/eval_baseline.sh GPUID refcoco unc