Alpha Matting Benchmark

Where the foreground object is opaque.

Alpha matting benchmark.

Technical report tr 188 2 2009 03 may 2009. Even in regions where the alpha matte is 1 i e. Videomatting project is the first public objective benchmark of video matting methods. Sota for image matting on composition 1k mse metric include the markdown at the top of your github readme md file to showcase the performance of the model.

Cutting out an object and estimating its opacity mask known as image matting is a key task in many image editing applications. In this work we present an approach aimed to overcome these limitations. Badges are live and will be dynamically updated with the latest ranking of this paper. Official repository for the paper f b alpha matting this paper and project is under heavy revision for peer reviewed publication and so i will not be able to release the training code yet.

However most of the existing networks only predict the alpha matte and post processing methods must then be used to recover the original. A perceptually motivated online benchmark for image matting. Download the 16bit linear rgb fgd input images here be aware that the filesize is approximately 2gb. Submission page is now open.

It is based on a high quality ground truth alpha sequences collected using both capturing in front of a green plate and stop motion sequential photography we believe that our work will help ranking existing methods and challenge authors of new methods. Rest using a similar method to 25. In computer vision and pattern recognition 2009. While our method is primarily designed as a standalone matting tool we show that it can also be used for regularizing mattes obtained by sampling based methods.

Our method uses deep learning to di rectly compute the alpha matte given an input image and trimap. The rst image is from the alpha matting benchmark and the second image is from our 1000 testing images. Deep learning approaches have made significant progress by adapting the encoder decoder architecture of segmentation networks.

Christoph rhemann carsten rother jue wang margrit gelautz pushmeet kohli and pamela rott. This work was supported in part by microsoft research cambridge through its phd scholarship programme and a travel sponsorship. Ieee conference on pages 1826 1833. Our evaluation using the alpha matting benchmark suggests a significant performance improvement over the current methods.