| CARVIEW |
Order-aware Interactive Segmentation
3University of Central Florida 4University of North Carolina
Abstract
Interactive segmentation aims to accurately segment target objects with minimal user interactions. However, current methods often fail to accurately separate target objects from the background, due to a limited understanding of order, the relative depth between objects in a scene. To address this issue, we propose OIS: order-aware interactive segmentation, where we explicitly encode the relative depth between objects into order maps. We introduce a novel order-aware attention, where the order maps seamlessly guide the user interactions (in the form of clicks) to attend to the image features. We further present an object-aware attention module to incorporate a strong object-level understanding to better differentiate objects with similar order. Our approach allows both dense and sparse integration of user clicks, enhancing both accuracy and efficiency as compared to prior works. Experimental results demonstrate that OIS achieves state-of-the-art performance, improving mIoU after one click by 7.61 on the HQSeg44K dataset and 1.32 on the DAVIS dataset as compared to the previous state-of-the-art SegNext, while also doubling inference speed compared to current leading methods.
Methods
Each object has a specific order, or relative depth from each other, corresponding to its location in 3D space. By calculating the relative depth between each point in the image and the user prompt's location, we create order map, to effectively separate the prompt-selected object with others.
(Red dots are positive clicks, blue dots are negative clicks, darker means closer to prompt-selected object, while lighter areas are farther to prompt-selected object)
Demo Videos
Gate
Moto
Four Consecutive Images
Elephant
Results
(1) Multi-clicks result comparison of SegNext (current SOTA) and Ours:
(2) One click result comparison of SegNext (current SOTA) and Ours:
BibTeX
@article{wang2024order,
title={Order-Aware Interactive Segmentation},
author={Wang, Bin and Choudhuri, Anwesa and Zheng, Meng and Gao, Zhongpai and Planche, Benjamin and Deng, Andong and Liu, Qin and Chen, Terrence and Bagci, Ulas and Wu, Ziyan},
journal={arXiv preprint arXiv:2410.12214},
year={2024}
}