hduwbf / WSDVC

Proposal Generation via Distillation Learning

Teacher Networks

Run the bmn.sh files in the action/activity/highlight_teacher folders separately

sh bmn.sh

and move the checkpoints into event_student/checkpoint.

Student Networks

Run the bmn.sh file

sh bmn.sh

Cross-modal Matching

Training:

python train.py

Testing:

python eval.py config/anet_coot.yaml provided_models/anet_coot_AB.pth

Make pseudo dataset:

python make_ws_data.py config/anet_coot.yaml provided_models/anet_coot_AB.pth

Caption Generation

Training:

python mm_train.py

Testing:

sh mm_eval.sh