Motion-Guided-CRN

This is a PyTorch implementation of Cascaded Refinement Network described in

Ping Hu, Gang Wang, Xiangfei Kong , Jason Kuen, Yap-Peng Tan;. "Motion-Guided Cascaded Refinement Network for Video Object Segmentation." CVPR, 2018

Results on DAVIS2016 can be download here

Prerequisites

1. PyTorch
2. Opencv

Pretraining on Pascal_VOC

1. Edit img_path, gt_path, list_path in the file ./CRN_Pascal_Pretrain/config/CRN_Pascal.cfg.
2. Edit file ./CRN_Pascal_Pretrain/train.sh.
3. Run sh train.sh.

Pretraining on the DAVIS16 Training Split

1. Edit img_path, gt_path, list_path in the file ./CRN_DAVIS16_Pretrain/config/CRN_DAVIS16.cfg.
2. Copy a Pascal-pretrained model to ./CRN_DAVIS16_Pretrain/trained_model/.
3. Edit file ./CRN_DAVIS16_Pretrain/train.sh.
4. Run sh train.sh.

Online Finetuning

1. Edit img_path, gt_path, list_path in the file ./CRN_DAVIS16_Oneshot/config/CRN_DAVIS16.cfg.
2. Copy a DAVIS16-pretrained model to ./CRN_DAVIS16_Oneshot/trained_model/.
3. Edit file ./CRN_DAVIS16_Pretrain/train.sh.
4. Run sh train.sh.

Bibtex

@InProceedings{Hu_2018_CVPR,

author = {Hu, Ping and Wang, Gang and Kong , Xiangfei and Kuen, Jason and Tan, Yap-Peng},

title = {Motion-Guided Cascaded Refinement Network for Video Object Segmentation},

booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},

year = {2018}

}