Highly Efficient Natural Image Matting

10/25/2021
by   Yijie Zhong, et al.
0

Over the last few years, deep learning based approaches have achieved outstanding improvements in natural image matting. However, there are still two drawbacks that impede the widespread application of image matting: the reliance on user-provided trimaps and the heavy model sizes. In this paper, we propose a trimap-free natural image matting method with a lightweight model. With a lightweight basic convolution block, we build a two-stages framework: Segmentation Network (SN) is designed to capture sufficient semantics and classify the pixels into unknown, foreground and background regions; Matting Refine Network (MRN) aims at capturing detailed texture information and regressing accurate alpha values. With the proposed cross-level fusion Module (CFM), SN can efficiently utilize multi-scale features with less computational cost. Efficient non-local attention module (ENA) in MRN can efficiently model the relevance between different pixels and help regress high-quality alpha values. Utilizing these techniques, we construct an extremely light-weighted model, which achieves comparable performance with  1% parameters (344k) of large models on popular natural image matting benchmarks.

READ FULL TEXT

page 2

page 4

page 6

page 8

page 10

research
08/18/2021

Towards Deep and Efficient: A Deep Siamese Self-Attention Fully Efficient Convolutional Network for Change Detection in VHR Images

Recently, FCNs have attracted widespread attention in the CD field. In p...
research
02/14/2022

Convolutional Neural Network with Convolutional Block Attention Module for Finger Vein Recognition

Convolutional neural networks have become a popular research in the fiel...
research
03/31/2021

Human Perception Modeling for Automatic Natural Image Matting

Natural image matting aims to precisely separate foreground objects from...
research
01/14/2021

Image deblurring based on lightweight multi-information fusion network

Recently, deep learning based image deblurring has been well developed. ...
research
07/11/2023

Compact Twice Fusion Network for Edge Detection

The significance of multi-scale features has been gradually recognized b...
research
08/21/2022

LWA-HAND: Lightweight Attention Hand for Interacting Hand Reconstruction

Hand reconstruction has achieved great success in real-time applications...
research
07/28/2022

Real Image Restoration via Structure-preserving Complementarity Attention

Since convolutional neural networks perform well in learning generalizab...

Please sign up or login with your details

Forgot password? Click here to reset