Inside-Outside-Guidance (IOG)

This project hosts the code for the IOG algorithms for interactive segmentation.

Interactive Object Segmentation with Inside-Outside Guidance
Shiyin Zhang, Jun Hao Liew, Yunchao Wei, Shikui Wei, Yao Zhao

Updates:

• 2021.4.6 Create the interactive refinement branch for IOG.

Abstract

This paper explores how to harvest precise object segmentation masks while minimizing the human interaction cost. To achieve this, we propose an Inside-Outside Guidance (IOG) approach in this work. Concretely, we leverage an inside point that is clicked near the object center and two outside points at the symmetrical corner locations (top-left and bottom-right or top-right and bottom-left) of a tight bounding box that encloses the target object. This results in a total of one foreground click and four background clicks for segmentation. Our IOG not only achieves state-of-the-art performance on several popular benchmarks, but also demonstrates strong generalization capability across different domains such as street scenes, aerial imagery and medical images, without fine-tuning. In addition, we also propose a simple two-stage solution that enables our IOG to produce high quality instance segmentation masks from existing datasets with off-the-shelf bounding boxes such as ImageNet and Open Images, demonstrating the superiority of our IOG as an annotation tool.

Demo

IOG(3 points)	IOG(Refinement)	IOG(Cross domain)

Installation

1. Install requirement

• PyTorch = 0.4
• python >= 3.5
• torchvision = 0.2
• pycocotools

2. Usage You can start training with the following commands:

# training step
python train.py
python train_refinement.py

# testing step
python test.py
python test_refinement.py

# train step
python eval.py
python eval_refinement.py

We set the paths of PASCAL/SBD dataset and pretrained model in mypath.py.

Pretrained models

Network	Dataset	Backbone	Download Link
IOG	PASCAL + SBD	ResNet-101	IOG_PASCAL_SBD.pth
IOG	PASCAL	ResNet-101	IOG_PASCAL.pth
IOG-Refinement	PASCAL + SBD	ResNet-101	IOG_PASCAL_SBD_REFINEMENT.pth

Dataset

With the annotated bounding boxes (∼0.615M) of ILSVRCLOC, we apply our IOG to collect their pixel-level annotations, named Pixel-ImageNet, which are publicly available at https://github.com/shiyinzhang/Pixel-ImageNet.

Citations

Please consider citing our papers in your publications if it helps your research. The following is a BibTeX reference.

@inproceedings{zhang2020interactive,
  title={Interactive Object Segmentation With Inside-Outside Guidance},
  author={Zhang, Shiyin and Liew, Jun Hao and Wei, Yunchao and Wei, Shikui and Zhao, Yao},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={12234--12244},
  year={2020}
}