GitHub - insomnia94/ARG: Adaptive ROI Generation for Video Object Segmentation Using Reinforcement Learning

Prerequisites

Please refer to OnAVOS to prepare related dataset and install the environment of the benchmark method.

Train ARG on DAVIS 2016:

python ./train.py

Evaluate ARG on DAVIS 2016:

python ./eval.py configs/DAVIS16_online

Thanks for the work of Voigtlaender Paul. Our code is based on the implementation of OnAVOS.

Name		Name	Last commit message	Last commit date
Latest commit History 23 Commits
.idea		.idea
Forwarding		Forwarding
__pycache__		__pycache__
configs		configs
crf		crf
datasets		datasets
Config.py		Config.py
Constants.py		Constants.py
Engine.py		Engine.py
Log.py		Log.py
Measures.py		Measures.py
Measures.py~		Measures.py~
Network.py		Network.py
NetworkLayers.py		NetworkLayers.py
NetworkOutputLayers.py		NetworkOutputLayers.py
README.md		README.md
Trainer.py		Trainer.py
Util.py		Util.py
Util_Network.py		Util_Network.py
actor.py		actor.py
caffe_classes.py		caffe_classes.py
critic.py		critic.py
eval.py		eval.py
negative_actor.py		negative_actor.py
negative_critic.py		negative_critic.py
new_vgg19.py		new_vgg19.py
parameter.py		parameter.py
positive_actor.py		positive_actor.py
positive_critic.py		positive_critic.py
train.py		train.py