Code for ECCV 2020 spotlight paper: Video Object Segmentation with Episodic Graph Memory Networks
-
Install python (3.6.5), pytorch (version:1.0.1) and requirements in the requirements.txt files. Download the DAVIS-2017 dataset.
-
Download the pretrained model from googledrive and put it into the workspace_STM_alpha files.
-
Run 'run_graph_memory_test.sh' and change the davis dataset path, pretrainde model path and result path and the paths in local_config.py.
The segmentation results can be download from googledrive.
- DAVIS ( Val 2017):
In the inference stage, we ran using the default size of DAVIS (480p).
Mean J&F | J score | F score |
---|---|---|
82.8 | 80.0 | 85.2 |
- YouTube-VOS (Val 2018):
J Seen | F Seen | J Unseen | F Unseen | Mean |
---|---|---|---|---|
80.7 | 85.1 | 74.0 | 80.9 | 80.2 |
- DAVIS-2016:
J score | F score | Mean T |
---|---|---|
82.5 | 81.2 | 19.8 |
- Youtube-Objects:
Airplane | Bird | Boat | Car | Cat | Cow | Dog | Horse | Motorbike | Train | Mean |
---|---|---|---|---|---|---|---|---|---|---|
86.1 | 75.7 | 68.6 | 82.4 | 65.9 | 70.5 | 77.1 | 72.2 | 63.8 | 47.8 | 71.4 |
If you find the code and dataset useful in your research, please consider citing:
@inproceedings{lu2020video,
title={Video Object Segmentation with Episodic Graph Memory Networks},
author={Lu, Xiankai and Wang, Wenguan and Martin, Danelljan and Zhou, Tianfei and Shen, Jianbing and Luc, Van Gool},
booktitle={ECCV},
year={2020}
}
- Zero-shot Video Object Segmentation via Attentive Graph Neural Networks, ICCV 2019 (https://github.com/carrierlxk/AGNN)
- Video object segmentation using space-time memory networks, ICCV 2019 (https://github.com/seoungwugoh/STM)
- A Generative Appearance Model for End-to-End Video Object Segmentation, CVPR2019 (https://github.com/joakimjohnander/agame-vos)
- https://github.com/lyxok1/STM-Training
Any comments, please email: carrierlxk@gmail.com