Active Matting

NeurIPS 2018 (formerly known as NIPS)

Xin Yang*,1,2

Ke Xu*,1,2

Shaozhe Chen1

Shengfeng He3

Baocai Yin1

Rynson W.H. Lau2

(* joint first authors)

1 Dailian University of Technology

2 City University of Hong Kong

3 South China University of Technology

 

 

Abstract: Image matting is an ill-posed problem. It requires a user input trimap or some strokes to obtain an alpha matte of the foreground object. A fine user input is essential to obtain a good result, which is either time consuming or suitable for experienced users who know where to place the strokes. In this paper, we explore the intrinsic relationship between the user input and the matting algorithm to address the problem of where and when the user should provide the input. Our aim is to discover the most informative sequence of regions for user input in order to produce a good alpha matte with minimum labeling efforts. To this end, we propose an active matting method with recurrent reinforcement learning. The proposed framework involves human in the loop by sequentially detecting informative regions for trivial human judgement. Comparing to traditional matting algorithms, the proposed framework requires much less efforts, and can produce satisfactory results with just 10 regions. Through extensive experiments, we show that the proposed model reduces user efforts significantly and achieves comparable performance to dense trimaps in a user-friendly manner. We further show that the learned informative knowledge can be generalized across different matting algorithms.

 

 

Paper

PDF

Supplemental

PDF

Video

https://youtu.be/QcCBvS79fNE

Poster

PDF

Dataset

rendered-100

BibTeX

@INPROCEEDINGS{ActiveMatting:2018,

Author = {Xin Yang and Ke Xu and Shaozhe Chen and Shengfeng He and Baocai Yin and Rynson W.H. Lau},

Title = {Active Matting},

Booktitle = {Proc. NIPS},

Year = {2018}

}

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Last updated in November 2018