Alpha Matting Evaluation Website
www.alphamatting.com
Hide Help
How to use this page?
  1. Please be patient until all images have loaded completely.
  2. Move the mouse over the numbers in the table to see the corresponding images.
  3. Drag the red rectangle in the leftmost image to change the location of the zoom.
  4. Press and hold any key to temporarily deactivate the links.

Image matting evaluation results      Competition:   Low resolution  High resolution  
Error type:  SAD   MSE   Gradient   Connectivity  
Sum of Absolute Differences overall
avg.
small
avg.
large
avg.
user
Troll
(Strongly Transparent)
Input  
Doll
(Strongly Transparent)
Input  
Donkey
(Medium Transparent)
Input  
Elephant
(Medium Transparent)
Input  
Plant
(Little Transparent)
Input  
Pineapple
(Little Transparent)
Input  
Plastic bag
(Highly Transparent)
Input  
Net
(Highly Transparent)
Input  
  rank rank rank rank small large user small large user small large user small large user small large user small large user small large user small large user
TMFNet3.42.42.956.1 1 6.5 1 8.3 3 4.2 2 4.5 4 5 4 2.6 3 2.7 3 2.4 4 0.8 1 0.8 2 1.3 6 4 2 5.2 4 6.6 6 1.7 2 1.9 3 2.4 4 14.4 1 14.6 1 14.3 1 17.5 7 18 5 21.7 12
IamAlpha4.55.344.38 5 8.4 4 8.4 4 4.4 6 4.4 3 4.4 1 2.7 4 2.7 4 2.7 7 0.9 5 0.9 3 0.9 1 4.8 9 5.2 5 6.2 4 2.5 10 2.7 9 3.1 10 15.8 2 15.8 2 15.8 6 15.9 1 16.5 2 16.5 1
LFPNet4.84.13.66.87.6 4 8.1 3 9 8 4.3 5 4.3 2 5.1 6 3 11 3 9 2.8 11 0.8 2 0.8 1 1.5 10 3.9 1 4.2 1 5.3 1 1.7 1 1.7 1 2.2 1 17 5 18 9 17.9 15 16.6 4 16.6 3 17.6 2
TIMI-Net5.46.66.33.38.3 7 8.7 7 9 7 4.4 7 4.7 6 4.4 3 2.8 8 2.9 6 2 2 1 12 1.1 9 1.3 4 4.7 6 5.2 6 6.2 3 1.8 3 1.9 2 2.3 2 15.9 4 16.2 3 15.5 2 16.6 6 19.2 11 18 3
SIM6.46.95.56.88.3 6 8.7 6 9 6 4.8 16 4.8 12 6 19 2.2 1 2.2 1 2 1 0.9 8 0.9 6 1.1 2 4.7 7 5.1 3 6.3 5 2.2 6 2.3 6 2.5 5 15.9 3 16.3 4 16.3 8 17.8 8 18 6 20.9 8
RMat7.16.3697.5 3 7.5 2 7.9 2 3.7 1 3.8 1 5.3 10 2.8 5 2.9 7 2.3 3 0.9 6 0.9 4 1.4 8 4.2 3 4.3 2 5.5 2 1.9 4 2 4 2.4 3 20.5 26 20.9 27 22.4 37 16 2 16.2 1 20.6 7
PIIAMatting10.2712.6119.1 13 10 15 9.3 11 4.3 4 4.6 5 5.4 11 2.8 7 3.7 18 2.7 9 0.9 4 2.3 24 2 17 4.7 8 6.8 13 8.3 12 2.2 8 2.7 11 4.3 19 17.2 7 17.3 6 15.5 3 16.6 5 19.1 9 20.2 6
LSA Matting10.910.311.910.58.4 8 9 8 8.8 5 4.5 9 5.7 22 5 5 3.3 15 3.5 13 3.1 15 1 9 1 8 1.6 11 5.4 15 7.1 21 8.5 15 2.7 15 3 12 3.3 12 17.3 8 17.3 7 16.8 10 16.3 3 17.4 4 21.6 11
HDMatt1213.410.811.89.5 15 10 14 10.7 16 4.7 11 4.8 10 5.8 17 2.9 10 3 8 2.6 6 1.1 15 1.2 14 1.3 5 5.2 12 5.9 8 6.7 7 2.4 9 2.6 7 3.1 9 17.3 9 17.3 5 17 12 21.5 26 22.4 20 23.2 22
LiteMatting12.911.510.1178.6 9 9.1 9 9.2 9 4.6 10 4.7 7 5.6 14 3.5 19 3.7 15 3.2 19 1 11 1 7 2.6 31 4.5 4 5.3 7 7.8 9 2.1 5 2.2 5 3 7 18.8 19 19.4 21 19.5 26 20 15 19.2 10 22.8 21
TransMatting: Enhancing Transparent ...13.616.111.313.47.5 2 8.6 5 7.8 1 4.9 22 5 15 5.5 13 3.4 16 3.5 14 3.2 17 1.8 33 1.8 20 2.2 23 5.7 21 6.3 9 8.4 13 2.6 14 2.7 10 3 8 17.6 11 18 10 18.9 22 18.3 10 18.3 7 21.5 10
AdaMatting14.312.813.316.910.2 18 11.1 20 10.8 17 4.9 18 5.4 18 6.6 23 3.6 20 3.4 12 3.4 23 0.9 3 0.9 5 1.8 13 4.7 5 6.8 11 9.3 23 2.2 7 2.6 8 3.3 11 19.2 22 19.8 24 18.7 21 17.8 9 19.1 8 18.6 4
A2U Matting14.413.111.818.49.3 14 9.7 12 10.9 18 4.8 14 4.9 13 5.3 9 3 12 3.1 10 2.8 10 1 13 1.1 10 1.4 9 5.1 10 6.7 10 8.5 14 2.5 12 3 13 5.9 30 17.3 10 18.4 12 18.1 16 20.6 20 20.6 14 27.3 41
SampleNet Matting1511.815.317.99.1 12 9.7 13 9.8 13 4.3 3 4.8 9 5.1 7 3.4 18 3.7 19 3.2 20 0.9 7 1.1 12 2 15 5.1 11 6.8 12 9.7 26 2.5 11 4 20 3.7 14 18.6 18 19.3 20 19.1 25 20 14 21.6 17 23.2 23
FGI Matting1518.114.812.110.3 20 10.4 17 11.7 22 4.9 21 5.3 17 5.2 8 2.5 2 2.7 2 2.4 5 1.2 19 1.2 15 1.2 3 5.6 18 6.9 14 7.8 10 2.6 13 3.1 14 3 6 18.9 21 18.9 14 20 30 22.3 31 23.6 25 21.7 13
GCA Matting16.517.114.517.88.8 10 9.5 11 11.1 20 4.9 17 4.8 11 5.8 18 3.4 17 3.7 17 3.2 18 1.1 18 1.2 13 1.3 7 5.7 20 6.9 15 7.6 8 2.8 17 3.1 15 4.5 22 18.3 15 19.2 16 18.5 19 20.8 23 21.7 18 24.7 30
ATNet Matting17.819.417179.7 17 10.2 16 9.6 12 4.5 8 4.8 8 4.4 2 3.7 21 3.9 20 3.4 24 1.7 30 1.9 21 2.2 22 5.9 26 7 17 9.3 22 2.9 18 3.3 17 3.9 16 19.8 24 20.6 25 20.7 33 19 11 19.2 12 20.1 5
VDRN Matting1820.118.315.88.9 11 9.4 10 9.3 10 5.2 23 5.6 21 6.6 22 2.8 9 3.3 11 2.7 8 1.8 34 1.9 22 2 14 5.7 19 7.1 18 8.3 11 3 19 3.5 18 3.6 13 17.6 12 18.3 11 16.9 11 23.2 34 25.9 35 26.5 37
Deep Matting18.519.418.117.910.7 22 11.2 22 11 19 4.8 15 5.8 23 5.6 15 2.8 6 2.9 5 2.9 12 1.1 14 1.1 11 2 16 6 31 7.1 19 8.9 18 2.7 16 3.2 16 3.9 15 19.2 23 19.6 23 18.7 20 21.8 28 23.9 26 24.1 28
Information-flow matting20.621.821.918.310.3 19 11.2 21 12.5 24 5.6 29 7.3 27 7.3 28 3.8 22 4.1 22 3 14 1.4 21 2.3 23 2 19 5.9 28 7.1 20 8.6 16 3.6 23 5.7 24 4.6 23 18.3 14 19.3 19 15.8 5 20.2 18 22.2 19 22.3 17
IndexNet Matting21.723.620.820.612.6 39 13.4 24 11.4 21 4.8 13 4.9 14 5.7 16 3.3 14 4 21 3 13 1.1 16 1.5 17 1.6 12 6.4 35 7.5 24 8.9 19 3.4 20 4 19 4.1 17 18.6 17 19.1 15 18.5 18 23.4 35 25.1 32 29.3 49
DCNN Matting23.125.121.622.612 32 14.1 27 14.5 31 5.3 24 6.4 24 6.8 25 3.9 25 4.5 25 3.4 22 1.6 27 2.5 25 2.2 25 6 30 6.9 16 9.1 20 4 26 6 25 5.3 25 19.9 25 19.2 18 19.1 24 19.4 12 20 13 21.2 9
AlphaGAN23.524.124.422.19.6 16 10.7 18 10.4 15 4.7 12 5.3 16 5.4 12 3.1 13 3.7 16 3.1 16 1.1 17 1.3 16 2 18 6.4 37 8.3 33 9.3 24 3.6 22 5 21 4.3 20 20.8 27 21.5 28 20.6 32 25.7 49 28.7 47 26.7 40
CDI-Net2626.520.930.612.6 37 13.7 25 14 29 4.9 20 5.5 19 6.6 24 3.9 24 4.3 23 3.7 28 1.3 20 1.5 18 2.2 21 6.8 42 8.1 30 9.9 27 4.3 32 5.6 23 8.1 45 18.9 20 18.7 13 18.4 17 20.2 17 21.5 16 36.1 54
Context-aware Matting26.130.124.423.810.4 21 11.1 19 10.1 14 6.4 39 7.4 30 7.1 26 4.1 26 4.5 26 3.8 31 2.3 49 3.1 27 3 42 7.1 45 8.2 31 9.1 21 3.5 21 5.5 22 4.1 18 18.3 16 19.2 17 16.5 9 21.1 24 23.3 23 24.6 29
Three-layer graph matting28.423.626.53510.7 23 15.2 28 13.8 27 4.9 19 5.6 20 8.1 40 3.9 23 4.4 24 3.6 27 1 10 1.8 19 3 40 5.9 25 7.3 23 12.4 40 4.2 29 8 32 8.5 49 24.2 41 25.6 42 24.2 42 20.5 19 23.5 24 22.2 15
ATPM Matting30.333.83324.314 50 17.8 35 13.4 25 5.5 25 6.4 25 7.3 29 5.4 57 6.4 52 4.3 45 1.7 31 3.3 32 2.3 28 6.8 41 8 27 8.7 17 4.2 27 7.5 29 5.5 26 17.2 6 17.6 8 15.7 4 22.6 33 37.3 56 22.8 20
Three Stages Matting3231.332.632.111.7 31 13.9 26 13.9 28 5.6 26 7.4 29 7.9 34 4.6 38 5.5 39 4.2 40 2.2 45 4 39 3.1 45 6.5 38 11 50 11.9 35 4 24 6.5 26 4.5 21 23.3 35 23.2 37 22.3 36 19.6 13 20.8 15 22.4 18
LNSP Matting33.529.934.336.412.2 33 22.5 53 19.5 58 5.6 27 8.1 33 8.8 50 4.6 36 5.9 44 3.6 26 1.5 24 3.5 35 3.1 44 6.2 32 8.1 29 10.7 31 4 25 7.1 27 6.4 32 21.5 30 20.8 26 16.3 7 22.5 32 24.4 27 27.8 43
CSC Matting33.737.130.133.913.6 47 15.6 29 14.5 30 6.2 37 7.5 31 8.1 41 4.6 39 4.8 28 4.2 43 1.8 36 2.7 26 2.5 30 5.5 17 7.3 22 9.7 25 4.6 34 7.6 30 6.9 36 23.7 37 23 35 21 34 26.3 50 27.2 40 25.2 32
Graph-based sparse matting34.835.135.633.512.6 40 20.5 45 14.8 35 5.7 31 7.3 26 6.4 21 4.5 34 5.3 35 3.7 29 1.4 23 3.3 33 2.3 27 6.3 34 7.9 25 11.1 32 4.2 28 8.3 35 6.4 31 28.7 54 31.3 55 27.1 51 23.6 37 25.1 31 27.3 42
KL-Divergence Based Sparse Sampling34.933.135.136.411.6 30 17.5 34 14.7 32 5.6 28 8.5 38 8 36 4.9 50 5.3 33 3.7 30 1.5 25 3.5 34 2.1 20 5.8 23 8.3 32 14.1 48 5.6 45 9.3 40 8 44 24.6 42 27.7 49 28.9 54 20.7 22 22.7 21 23.9 27
Patch-based Matting3528.837.538.610.9 25 19 39 15.7 41 6 34 9.5 47 8.3 43 4.3 30 5.2 32 4.2 42 1.6 26 3.2 30 2.6 32 5.2 13 9 38 12.4 37 4.7 35 9.7 42 7 37 21.6 31 21.7 29 24.9 44 23.5 36 28.1 43 25.6 33
TSPS-RV Matting36.334.436.538.111.3 29 16.4 32 13.7 26 6.1 35 8.1 35 8.6 47 4.5 32 5.4 37 4.1 39 1.4 22 3.3 31 3.5 54 7.9 53 8.9 36 12.4 39 6.2 49 9 38 8.7 51 22.8 34 23.5 38 21.4 35 20.7 21 28.5 45 22.2 14
Iterative Transductive Matting3737.937.435.813.1 44 17.2 33 15.6 40 5.7 30 8.6 40 7.8 33 5.1 53 5.5 38 3.9 32 1.9 37 5.8 50 2.6 33 6.6 39 8.5 34 13.8 47 5.4 40 10 43 7.4 41 25.5 44 24 39 23.8 41 20.1 16 22.7 22 22.7 19
SVR Matting37.440.438.333.518.7 62 30.7 64 19.1 55 6.8 49 7.7 32 7.6 32 4.7 45 5 31 3.4 21 1.9 38 4.7 42 2.9 38 5.8 22 8.7 35 10.5 28 4.3 30 8 33 5.6 28 21.2 29 22.1 32 17.1 14 25.6 48 26.1 37 30.6 52
Comprehensive sampling37.533.938.340.311.2 28 18.5 38 14.8 33 6.5 43 9.5 46 8.9 52 4.5 33 4.9 29 4.1 38 1.7 29 3.1 29 2.3 29 5.4 16 9.8 42 13.4 45 5.5 43 11.5 49 7.4 43 23.9 39 22 31 22.8 38 23.8 40 28 42 28.1 44
Comprehensive Weighted Color and Texture383739.637.314.6 51 16 30 15.7 42 6.8 48 10 50 7.9 35 4.3 29 5 30 4.1 36 1.7 28 3.5 36 2.2 24 5.4 14 9.9 44 12.8 43 4.3 31 7.4 28 5.2 24 28.3 53 28.1 50 25.4 47 24 42 30.2 49 28.7 47
Sparse coded matting38.641.939.534.413.7 48 25.8 60 14.8 36 6.4 40 8.2 36 6.2 20 4.7 40 5.4 36 4 34 1.8 35 3.1 28 2.3 26 5.9 27 8 28 10.6 29 4.5 33 8 31 5.5 27 30.3 56 33.1 58 29.2 55 27.7 56 27.2 39 29 48
LocalSamplingAndKnnClassification40.242.938.439.312.6 38 16 31 12.4 23 5.8 32 8.1 34 8 37 4.5 35 5.5 40 4.1 35 2.2 46 5.1 45 3.4 52 8.1 54 10.5 46 15.6 53 7.3 53 12.3 52 9.4 52 24.1 40 21.8 30 19.7 27 24.7 45 24.8 29 25.9 35
Weighted Color and Texture Matting40.838.343.34113.1 45 17.8 36 15.8 43 6.5 42 9.4 44 8.6 46 4.2 28 4.7 27 3.9 33 1.7 32 6 51 2.7 34 6.4 36 11 49 16.3 54 4.8 36 9.1 39 6.5 33 23.7 36 24.8 41 23.2 39 26.5 51 40.2 59 28.5 46
LNCLM matting41.143.641.138.610.9 26 11.2 23 16.7 49 6.9 51 8.9 43 7.2 27 5.6 59 7 60 4.1 37 2.5 53 5.1 46 3.5 55 7.5 48 10 45 12 36 5.5 44 11.3 46 7.3 40 25.2 43 23 36 19.9 29 21.2 25 25 30 26.4 36
CCM41.244.441.33813.8 49 20.8 46 16.9 51 6.4 41 8.9 42 8.2 42 4.7 41 5.9 45 3.6 25 2.5 51 4.3 40 3 39 7 44 9 37 10.6 30 4.9 37 8.1 34 5.7 29 25.6 45 27.5 48 24.5 43 25.4 47 26.4 38 28.2 45
Shared Matting41.339.445.139.410.8 24 20.5 44 15 37 7.8 55 11.6 56 8.1 38 4.2 27 5.3 34 4.2 41 2.1 40 5.8 49 2.9 37 5.9 29 9.2 39 11.4 34 5 38 8.8 36 6.8 35 34.9 61 34.9 59 34.3 59 23.9 41 28.4 44 25.7 34
Global Sampling Matting43.840.346.644.410.9 27 22.7 54 15.4 38 6.3 38 9.5 48 9 53 4.7 42 6.4 54 4.3 47 2.2 47 5.6 48 3.4 51 6.9 43 9.6 41 12.9 44 6.3 50 12.5 53 8.6 50 25.8 46 27.5 47 25.3 46 22 29 24.4 28 23.7 26
SRLO Matting44.643.147.443.414.7 52 18 37 17.7 53 6.9 50 10.7 52 8.9 51 4.9 51 5.7 42 4.7 55 2.1 42 6.5 52 2.8 36 6.3 33 10.9 48 15.2 52 5.4 39 11.6 50 7 39 26.5 51 29.7 52 25.1 45 21.7 27 28.5 46 22.3 16
Segmentation-based matting4545.844.145.112.8 43 23.5 56 16.6 48 6.6 44 8.3 37 7.3 30 4.8 49 6.1 48 4.3 48 2.1 41 3.9 38 3.1 43 6.7 40 8 26 13.4 46 6 48 8.8 37 8.2 46 31.6 57 35.6 60 38.8 61 24.5 44 32 51 26.7 39
KNN Matting46.248.148.342.316.2 57 19.7 40 16.8 50 8 56 11 53 9 55 4.7 43 6.7 57 4.3 44 3 58 7.7 58 3.7 56 9.2 57 11.3 51 11.3 33 6 47 10.4 45 6.7 34 18.1 13 19.6 22 17 13 27.4 54 41 60 32.7 53
Improved color matting46.346.545.646.614.9 53 24.5 58 20 59 6.7 46 9.5 45 8.5 45 4.6 37 6.1 49 4.3 49 2.6 55 5.4 47 3.4 53 7.5 49 9.9 43 12.5 41 6 46 10.1 44 8.4 48 26.1 47 26.7 46 23.6 40 23.8 39 25.6 33 26.7 38
Local Spline Regression (LSR)46.648.843.447.812.2 34 20 42 16.2 44 6.1 36 8.8 41 8.1 39 5.2 55 6.2 51 4.6 52 2.2 48 4.9 44 3.1 46 9.4 59 11.9 53 18.3 57 8.2 56 11.4 47 10.1 54 26.4 49 22.5 33 20.2 31 27 53 26 36 39.8 59
Global Sampling Matting (filter version)46.844.449.646.412.3 35 24.3 57 16.3 45 7.3 53 10.2 51 9.5 56 5.1 54 6.4 53 4.7 56 2.4 50 4.7 43 3.2 48 5.9 24 9.5 40 12.4 38 6.5 51 13.1 54 8.3 47 26.5 50 28.3 51 30 57 23.6 38 29.1 48 23.5 24
Learning Based Matting4848.647.348.116 56 22 52 18.7 54 6.6 45 7.4 28 7.4 31 4.8 47 6.1 47 4.3 50 2.1 39 3.7 37 2.8 35 7.5 50 14.5 59 19.5 60 8.6 60 14.1 57 14.6 62 22.5 32 24.8 40 19.9 28 34.6 60 38.5 58 51.2 65
LMSPIR49.148.150.948.315.2 55 20 41 19.1 56 6.7 47 11.2 54 8.7 49 4.8 48 5.8 43 4.6 54 2.1 43 6.8 55 3 41 7.9 52 14.3 58 20.2 61 5.5 42 11.4 48 7 38 29.7 55 31 54 29.6 56 24 43 34.2 54 25 31
Shared Matting (Real Time)49.548.150.549.812.4 36 21.6 48 16.3 46 9.5 59 13.5 58 9.9 57 4.4 31 5.6 41 4.4 51 2.5 54 6.8 56 3.2 49 7.1 46 10.8 47 12.6 42 5.4 41 9.7 41 7.4 42 35.5 63 35.8 61 35.5 60 27.6 55 33.4 52 29.8 51
Closed-Form Matting5048.647.653.912.7 41 21.9 51 17.2 52 5.9 33 8.5 39 8.6 48 4.7 44 6 46 4.3 46 2.2 44 4.6 41 3.3 50 9.3 58 12.1 54 19.3 59 8.3 57 14.9 59 13.4 61 34.2 60 32.4 57 27.4 52 26.5 52 25.7 34 48.3 63
Improving Sampling Criterion53.752.855.552.812.7 42 21.1 47 16.4 47 11.5 62 17 63 12.1 62 5.8 61 8.1 63 5.7 61 5.4 66 12.1 65 6.6 65 9.6 61 15.7 60 17.8 55 10.2 62 17 60 11.2 59 23.8 38 26.7 45 25.9 48 22.2 30 27.3 41 23.7 25
Cell-based matting Laplacian54.356.45353.415.1 54 21.7 49 15.6 39 7.6 54 11.4 55 9 54 5.7 60 6.7 58 5.2 59 3.1 59 6.7 53 4.5 59 8.6 55 11.4 52 14.9 50 8.5 59 13.5 56 11.1 57 27.7 52 26.2 44 27.6 53 32.4 58 37.6 57 37.9 56
Large Kernel Matting54.556.153.953.517.2 58 21.8 50 20.7 60 7.2 52 9.6 49 8.4 44 5.3 56 6.6 56 4.6 53 2.9 57 8.2 60 4.2 57 8.6 56 12.1 55 14.7 49 8 55 13.4 55 11.2 58 33 58 31.8 56 26.1 49 32.1 57 32 50 38.4 58
Robust Matting55.251.556.15817.3 59 28.4 62 21.1 61 10.1 60 16.9 62 11.4 60 4.8 46 6.5 55 5 58 2.8 56 7.3 57 4.4 58 7.3 47 14 57 18.1 56 6.8 52 14.6 58 10.6 56 22.7 33 26.1 43 32.1 58 34.4 59 37 55 38 57
SPS matting57.256.660.354.813.4 46 23 55 14.8 34 12.5 63 18.5 64 12.4 63 6.6 65 9.4 66 6.5 66 4.5 64 12.7 66 5.4 62 9.5 60 16.4 62 18.6 58 9.8 61 18.1 63 10.6 55 26.1 48 30.9 53 26.2 50 25 46 33.5 53 29.5 50
High-res matting58.256.859.558.418.6 61 25.8 59 24.6 63 8.6 57 14.1 59 11.1 59 5 52 6.2 50 4.8 57 2.5 52 8.3 61 3.2 47 7.8 51 14 56 21.4 62 8.5 58 18.1 64 12.2 60 35.3 62 38.1 63 42.6 64 38.7 61 54.6 64 36.8 55
Transfusive Weights58.658.858.558.523.2 63 26.7 61 22.4 62 9.2 58 11.9 57 10.4 58 6.1 63 8.1 62 5.9 63 3.8 62 8 59 5.3 61 12.9 64 21.4 64 37.6 68 13.1 67 22.8 66 18.3 67 20.9 28 22.9 34 19 23 59.5 65 65.7 65 72.8 66
Random Walk Matting61.963.559.562.817.9 60 20.3 43 19.4 57 11.3 61 15.6 60 11.8 61 5.8 62 7 59 6.3 65 3.4 60 6.7 54 4.6 60 13.1 66 22.1 65 27.4 64 12.3 66 18 62 15.7 65 44.1 67 43.5 66 41 63 75.1 66 81.8 67 80.6 67
Geodesic Matting63.16462.962.526.9 66 38.5 67 32.5 66 14.2 64 16.5 61 17.4 65 11.7 68 14 69 9.4 68 7.6 68 15.1 67 8.7 68 12.8 63 16.7 63 15.1 51 7.3 54 12.1 51 9.8 53 37.3 65 37.4 62 42.8 65 48.6 64 50 63 48.6 64
Iterative BP Matting64.263.564.16523.6 64 29.9 63 27.2 64 16.7 65 24.3 66 20.7 68 6.7 66 9 64 6.3 64 3.8 61 11.3 63 6.8 67 14.1 67 22.8 66 27.9 65 11.4 64 19 65 14.7 63 33.4 59 39.3 64 47.5 68 40.6 62 48.1 62 45.1 61
Easy Matting64.6656464.923.9 65 32.6 65 30 65 17.1 66 21.8 65 19.4 67 6.3 64 7.5 61 5.8 62 4.7 65 10.5 62 5.6 63 12.1 62 15.7 61 22.9 63 11.2 63 17 61 14.8 64 49.5 68 49.6 68 46.2 67 77.8 67 108.6 69 109.2 68
Improved Bayesian65.16565.564.931.2 68 34.9 66 33.9 68 17.8 67 24.4 67 17.2 64 5.5 58 9.3 65 5.3 60 3.9 63 11.4 64 6.6 66 13 65 29.5 67 34.6 67 11.9 65 24.5 67 16.1 66 42.9 66 44.6 67 46 66 92.4 68 46.5 61 45.4 62
Bayesian Matting66.466.567.465.330.3 67 42.4 68 33.4 67 19.2 68 25.8 68 18.4 66 10.8 67 12.4 67 10.8 69 6.6 67 18.5 69 6.2 64 14.2 68 29.8 68 33.2 66 15.4 68 30.6 68 19.7 68 35.8 64 40.6 65 39.6 62 45.3 63 76.8 66 43.6 60
Poisson Matting68.86968.668.851.8 69 56.2 69 52 69 28.3 69 43.5 69 30.7 69 12.1 69 13.7 68 9.2 67 11.7 69 18.4 68 11.2 69 22.4 69 36.8 69 55.5 69 21.4 69 32.2 69 22.7 69 53.6 69 72.9 69 58.4 69 125.5 69 84.8 68 139.7 69
Move the mouse over the numbers in the table to see the corresponding images.Click to compare with the ground truth. Press a key to deactivate the links (to better use the zoom).

Troll - Input image
Loading zoom, please wait Drag the window to change the zoom.

References

MethodReference and notesImplementation details
Closed-Form MattingA. Levin, D. Lischinski, Y. Weiss, A Closed Form Solution to Natural Image Matting, CVPR, 2006Matlab implementation on a Intel Core2 Quad with 2.4 GHZ
Bayesian MattingY.Y. Chuang, B. Curless, D. Salesin, R. Szeliski, A Bayesian Approach to Digital Matting, CVPR, 2001C++ implementation on a Intel Core2 Quad with 2.4 GHZ
Poisson MattingJ. Sun, J. Jia, C.K. Tang, H.Y. Shum, Poisson matting, SIGGRAPH, 2004Matlab implementation on a Intel Core2 Quad with 2.4 GHZ
Easy MattingY. Guan, W. Cheny, X. Liang, Z. Ding, Q. Peng, Easy Matting: A Stroke Based Approach for Continuous Image Matting, Eurographics, 2006C++ implementation on a Intel Core2 Quad with 2.4 GHZ
Random Walk MattingL. Grady, T. Schiwietz, S. Aharon, Random Walks For Interactive Alpha-Matting, VIIP, 2005Matlab/C++ implementation on a Intel Core2 Quad with 2.4 GHZ
Robust MattingJ. Wang, M. Cohen, Optimized Color Sampling for Robust Matting, CVPR, 2007C++ implementation on a Intel Core2 Quad with 2.4 GHZ
Geodesic MattingXue Bai, Guillermo Sapiro, A geodesic framework for fast interactive image and video segmentation and matting, ICCV 2007C++ implementation on a Intel Core2 Duo with 2.53 GHZ
Iterative BP MattingJue Wang, Michael Cohen, Jue Wang and Michael F. Cohen. An iterative optimization approach for unified image segmentation and matting. ICCV 2005.c++ implementation on a Intel Core2 Quad with 3 GHZ
Improved color mattingC. Rhemann, C. Rother, M. Gelautz, Improving Color Modeling for Alpha Matting. BMVC, 2008Matlab implementation on a Intel Core2 Duo with 2.4 GHZ
High-res mattingC. Rhemann, C. Rother, A. Rav-Acha, M. Gelautz, T. Sharp, High ResolutionMatting via Interactive Trimap Segmentation. CVPR, 2008Matlab/C++ implementation on a Intel Core2 Duo with 2.4 GHZ
Large Kernel MattingKaiming He, Jian Sun, and Xiaoou Tang, Fast Matting using Large Kernel Matting Laplacian Matrices, CVPR 2010C++ implementation on a Intel Core Duo with 2 GHZ
Segmentation-based mattingChristoph Rhemann, Carsten Rother, Pushmeet Kohli, Margrit Gelautz, A Spatially Varying PSF-based Prior for Alpha Matting, CVPR 2010Matlab/C++ implementation on a Intel Core2 Quad with 2.39 GHZ
Shared MattingEduardo S. L. Gastal and Manuel M. Oliveira, Shared Sampling for Real-Time Alpha Matting, Eurographics, 2010C++/GLSL implementation on a Core 2 Quad with 2.8 GHZ
Shared Matting (Real Time)Eduardo S. L. Gastal and Manuel M. Oliveira, Shared Sampling for Real-Time Alpha Matting, Eurographics, 2010C++/GLSL implementation on a Core 2 Quad with 2.8 GHZ
Learning Based MattingYuanjie Zheng, Chandra Kambhamettu, Yuanjie Zheng, Chandra Kambhamettu. Learning Based Digital Matting. ICCV 2009. SOURE CODEMatlab/C++ implementation on a Intel Core2 Duo with 2.53 GHZ
LMSPIRBei He, Guijin Wang, Zhiwei Ruan, Xuanwu Yin, Xiaokang Pei, Xinggang Lin, Local Matting based on Sample-pair Propagation and Iterative Refinement, ICIP 2012C++ implementation on a Intel Core2 Dual with 2 GHZ
SVR MattingZhanpeng Zhang, Qingsong Zhu, Yaoqin Xie, Learning Based Alpha Matting using Support Vector Regression, ICIP 2012Matlab implementation on a Intel Pentium Dual-Core with 3 GHZ
Cell-based matting LaplacianChen-Yu Tseng and Sheng-Jyh Wang, A cell-based matting Laplacian for contrast enhancement, ICIP 2012C++ implementation on a Intel Core i3 with 3 GHZ
Global Sampling MattingKaiming He, Christoph Rhemann, Carsten Rother, Xiaoou Tang, and Jian Sun, A Global Sampling Method for Alpha Matting, CVPR 2011C++ implementation on a Intel Core2 with 2 GHZ
Global Sampling Matting (filter version)Kaiming He, Christoph Rhemann, Carsten Rother, Xiaoou Tang, and Jian Sun, A Global Sampling Method for Alpha Matting, CVPR 2011 (using the guided filter in "Guided Image Filtering", ECCV 2010, by Kaiming He, Jian Sun, and Xiaoou Tang )C++ implementation on a Intel Core2 with 2 GHZ
Local Spline Regression (LSR)Shiming Xiang, Feiping Nie, Changshui Zhang, Semi-Supervised Classification via Local Spline Regression. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 32, no. 11, pp. 2039-2053, 2010C++ implementation on a Intel Core2 with 3 GHZ
Weighted Color and Texture MattingE.Shahrian and D.Rajan, Weighted Color and Texture Sample Selection for Image Matting , CVPR 2012. matlab implementation on a Intel(R) Xeon(R) with 2.93 GHZ
KNN MattingQifeng Chen, Dingzeyu Li, Chi-Keung Tang, KNN Matting, CVPR 2012Matlab implementation on a Intel Core 2 Duo with 2.13 GHZ
SRLO MattingBei He, Guijin Wang, Xuanwu Yin, Bo Liu, Chenbo Shi, Xinggang Lin, High-accuracy and Quick Matting based on Sample-pair Refinement and Local Optimization, IEICE trans. on Information & Systems, 2013C++ implementation on a Intel Core2 Dual with 2 GHZ
LNSP MattingXiaowu Chen, Dongqing Zou, Ping Tan, Image Matting with Local and Nonlocal Smooth Priors, CVPR 2013matlab implementation on a intel core2 with 2.2 GHZ
CCMYongfang Shi, Au, O.C., Jiahao Pang, Tang, K., Wenxiu Sun, Hong Zhang, Wenjing Zhu, and Luheng Jia, Color Clustering Matting, ICME2013.Matlab implementation on a Intel Core i7 with 2.8 GHZ
Iterative Transductive MattingBei He, Guijin Wang, Chenbo Shi, Xuanwu Yin, Bo Liu, Xinggang Lin, Iterative Transductive Matting, ICIP 2013Matlab implementation on a Intel Core2 Dual with 2.2 GHZ
Improving Sampling CriterionJun Cheng, Zhenjiang Miao, Improving Sampling Criterion for Alpha Matting, RACVPR2013 in Conjunction with ACPR2013C++ implementation on a Core i5 with 2.5 GHZ
Transfusive WeightsKaan Yucer, Alexander Sorkine-Hornung, and Olga Sorkine-Hornung, Transfusive Weights for Content-Aware Image Manipulation, VMV2013Matlab implementation on a Quad-Core Intel Xeon with 3.2 GHZ
Comprehensive samplingE.Shahrian, D.Rajan, B.Price, S.Cohen, Improving Image Matting using Comprehensive Sampling Sets, CVPR 2013.Matlab implementation on a Intel Xeon with 2.4 GHZ
Comprehensive Weighted Color and TextureE.Shahrian , D.Rajan, Weighted Color and Texture Sample Selection for Image Matting, IEEE Transaction on Image Processing , Volume:PP, Issue: 99 , 2013. Matlab implementation on a Intel i7 with 3.5 GHZ
SPS mattingAhmad Al-Kabbany and Eric Dubois, "Improved global-sampling matting using sequential pair-selection strategy", In Proc. Visual Information Processing andCommunication V(SPIE), San Francisco, February 2014.Matlab implementation on a Intel Core2 Quad with 2.66 GHZ
Improved BayesianWenshuang Tan, Automatic Matting of Identification Photos, CAD/Graphics, 2013C++ implementation on a Intel Core(TM)i7-2600 with 3.4 GHZ
Sparse coded mattingJubin Johnson, Deepu Rajan, Hisham Cholakkal, Sparse Codes as Alpha Mattes, BMVC 2014.Matlab implementation on a Intel Xeon with 3.2 GHZ
LNCLM mattingB.-K. Kim, M. Jin, W.-J Song, Local and Nonlocal Color Line Models for Image Matting, IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, vol. E97-A. no.8, pp. 1814-1819, Aug. 2014.MATLAB implementation on a Intel Core i5-3570 with 3.4 GHZ
Graph-based sparse mattingJubin Johnson, Ehsan Shahrian Varnousfaderani, Hisham Cholakkal, and Deepu Rajan, Sparse Coding for Alpha Matting, IEEE Transactions on Image Processing, Volume: PP, Issue: 99, 2016.Matlab implementation on a Intel Xeon with 3.2 GHZ
KL-Divergence Based Sparse SamplingLevent Karacan, Aykut Erdem and Erkut Erdem, Image Matting with KL-Divergence Based Sparse Sampling,IEEE International Conference on Computer Vision(ICCV) 2015)Matlab implementation on a Intel® Xeon(R) CPU E5-2620 with 2 GHZ
LocalSamplingAndKnnClassificationXiao Chen, Fazhi He, A Sampling-Propagation Matting Method Based on the Sample Validity and KNN Classification Labeling, Journal of Computer-Aided Design and Computer Graphics (CADCG), vol. 28(12), pp. 2186-2194, 2016.matlab implementation on a intel i3 2120 with 3.3 GHZ
DCNN MattingDonghyeon Cho, Yu-Wing Tai, Inso Kweon, Natural Image Matting using Deep Convolutional Neural Networks. ECCV 2016matlab implementation on a Intel Core i7 with 3.4 GHZ
CSC MattingXiaoxue Feng, Xiaohui Liang, Zili Zhang, A Cluster Sampling Method for Image Matting via Sparse Coding. ECCV 2016 Matlab implementation on a Intel(R) Core(TM) i5-3470 with 3.2 GHZ
Patch-based MattingGuangying Cao, Jianwei Li, Zhiqiang He, Xiaowu Chen, Divide and Conquer: A Self-Adaptive Approach for High-Resolution Image Matting. International Conference on Virtual Reality and Visualization (ICVRV 2016)matlab implementation on a i7 with 3.6 GHZ
TSPS-RV MattingAhmad Al-Kabbany and Eric Dubois, Matting with Sequential Pair Selection Using Graph Transduction. The 21st International Symposium on Vision, Modeling, and Visualization (VMV 2016)Matlab implementation on a Intel Core2 Quad with 2.66 GHZ
Deep MattingNing Xu, Brian Price, Scott Cohen and Thomas Huang, Deep Image Matting, CVPR 2017Matlab implementation on a I7 with 2.7 GHZ
Information-flow mattingYagiz Aksoy, Tunc Ozan Aydin and Marc Pollefeys, Designing Effective Inter-Pixel Information Flow for Natural Image Matting, CVPR 2017Matlab implementation on a Intel Xeon with 3.5 GHZ
ATPM MattingXiangyu Zhu, Ping Wang, Zhenghai Huang, Adaptive Propagation Matting Based on Transparency of Image, Multimedia Tools and Applications, vol. 77, pp. 9089-19112, 2018, doi: 10.1007/s11042-017-5357-7matlab implementation on a Intel Xeon with 2.4 GHZ
Three-layer graph mattingChao Li, Ping Wang, Xiangyu Zhu, Huali Pi, Three-layer graph framework with the sumD feature for alpha matting. Computer Vision and Image Understanding, vol. 162, pp. 34-45, 2017c++, octave implementation on a Intel Xeon with 3.2 GHZ
Three Stages MattingXiao Chen, A Three-Stage Matting Method, IEEE Access, 5(99):27732-27739, 2017matlab implementation on a i7 6700k with 4 GHZ
AlphaGANSebastian Lutz, Konstantinos Amplianitis, Aljosa Smolic, AlphaGAN: Generative Adversarial Networks for Natural Image Matting, BMVC 2018Python implementation on a Intel i7-6700 with 3.4 GHZ
SampleNet MattingJingwei Tang, Yagiz Aksoy, Cengiz Oztireli, Markus Gross, Tunc Ozan Aydin, Learning-based Sampling for Natural Image Matting, CVPR 2019Python implementation on a Intel Core i7-7700K with 4.7 GHZ
VDRN MattingHuan Tang, Yujie Huang, Ming'e Jing, Yibo Fan, Xiaoyang Zeng, Very deep residual network for image matting, IEEE ICIP 2019 python implementation on a intel Xeon with 2.2 GHZ
AdaMattingShaofan Cai, Xiaoshuai Zhang, Haoqiang Fan, Haibin Huang, Jiangyu Liu, Jiaming Liu, Jiaying Liu, Jue Wang, and Jian Sun, Disentangled Image Matting, ICCV 2019 Python implementation on a Intel Core i7-7700K with 4.7 GHZ with 2.2 GHZ
IndexNet MattingHao Lu, Yutong Dai, Chunhua Shen, Songcen Xu, Indices Matter: Learning to Index for Deep Image Matting, ICCV 2019Python (PyTorch) implementation on a Intel i7-8700, GTX1070 with 3.2 GHZ
Context-aware MattingQiqi Hou, Feng Liu, Context-aware Image Matting for Simultaneous Foreground and Alpha Estimation. ICCV 2019Python, Tensorflow implementation on a 1080 Ti with 2.2 GHZ
GCA MattingYaoyi Li, Hongtao Lu, Natural Image Matting via Guided Contextual Attention, AAAI 2020python implementation on a Intel(R) Xeon(R) CPU E5-2640, GeForce RTX 2080 Ti with 2.6 GHZ
ATNet MattingF. Zhou, Y. Tian, Z. Qi, Attention Transfer Network For Nature Image Matting, IEEE Transactions on Circuits and Systems for Video Technology, doi: 10.1109/TCSVT.2020.3024213python implementation on a 1080Ti with 2.2 GHZ
PIIAMattingYuhao Liu, Jiake Xie, Yu Qiao, Yong Tang, Xin Yang, Prior-Induced Information Alignment for Image Matting, IEEE Transactions on Multimedia, doi: 10.1109/TMM.2021.3087007Python implementation on a GTX 2080ti with 2.6 GHZ
SIMYanan Sun, Chi-Keung Tang, Yu-Wing Tai, Semantic Image Matting, CVPR 2021python implementation on a GTX 2080ti with 2.6 GHZ
HDMattHaichao Yu, Ning Xu, Zilong Huang, Yuqian Zhou, Humphrey Shi, High-Resolution Deep Image Matting, AAAI 2021Python implementation on a Tesla V100 with 3.2 GHZ
TIMI-NetYuhao Liu, Jiake Xie, Xiao Shi, Yu Qiao, Yujie Huang, Yong Tang, Xin Yang, Tripartite Information Mining and Integration for Image Matting, ICCV 2021python implementation on a tesla v100 with 3.5 GHZ
A2U MattingYutong Dai, Hao Lu, Chunhua Shen, Learning Affinity-Aware Upsampling for Deep Image Matting, CVPR 2021Python implementation on a GTX 1080 Ti with 3.2 GHZ
LFPNetQinglin Liu, Haozhe Xie, Shengping Zhang, Bineng Zhong, Rongrong Ji, Long-Range Feature Propagating for Natural Image Matting, ACM MM 2021Python implementation on a Nvidia GTX 1080Ti with 1.5 GHZ
IamAlphaAvinav Goel, Manoj Kumar, Pavan Sudheendra, IamAlpha: Instant and Adaptive Mobile Network for Alpha Matting, BMVC 2021Python implementation on a Nvidia Tesla P40 with 3.1 GHZ
TMFNetanonymous, Trimap-guided Feature Mining and Fusion Network for Natural Image Matting, submission to CVIU, 2022python implementation on a Tesla V100 with 3.5 GHZ
FGI MattingHang Cheng, Shugong Xu, Xiufeng Jiang, Rongrong Wang, Deep Image Matting with Flexible Guidance Input, BMVC 2021python implementation on a RTX 2080 Ti with 2.6 GHZ
LSA MattingRui Wang, Jun Xie, Jiacheng Han, Dezhen Qi, Improving Deep Image Matting via Local Smoothness Assumption, IEEE ICME 2022Python implementation on a Intel Xeon with 2.3 GHZ
RMatYutong Dai, Brian Price, He Zhang, Chunhua Shen, Boosting Robustness of Image Matting with Context Assembling and Strong Data Augmentation, CVPR 2022python implementation on a Tesla V100 with 3.2 GHZ
TransMatting: Enhancing Transparent ...Huanqia Cai, Fanglei Xue, Lele Xu, Lili Guo, TransMatting: Enhancing Transparent Objects Matting with Transformers, ECCV 2022, acceptedPython implementation on a Intel(R) Xeon(R) Silver 4210R with 2.4 GHZ
LiteMattinganonymous, Lightweight Image Matting via Efficient Non-Local Guidance, anonymous submission 2022python implementation on a Intel with 3.8 GHZ
CDI-NetZhiwei Ma, Guilin Yao, submission to Journal of Visual Communication and Image Representation, 2022Python implementation on a AMD Ryzen 9 5900HX with 3.3 GHZ