# Training Matting Models without Alpha Labels

arXiv: https://arxiv.org/abs/2408.10539 · HTML: https://mlsanity.com/paper/2408.10539/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 43 references that are in the map.
- First posted on arXiv 20 Aug 2024.

## Builds on (most cited first, 23 of 23)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Fully Convolutional Networks for Semantic Segmentation](https://mlsanity.com/paper/1411.4038/) (2014)
- [Unified Perceptual Parsing for Scene Understanding](https://mlsanity.com/paper/1807.10221/) (2018)
- [Panoptic Feature Pyramid Networks](https://mlsanity.com/paper/1901.02446/) (2019)
- [Context-aware Synthesis for Video Frame Interpolation](https://mlsanity.com/paper/1803.10967/) (2018)
- [Deep Image Matting](https://mlsanity.com/paper/1703.03872/) (2017)
- [MODNet: Real-Time Trimap-Free Portrait Matting via Objective Decomposition](https://mlsanity.com/paper/2011.11961/) (2020)
- [Indices Matter: Learning to Index for Deep Image Matting](https://mlsanity.com/paper/1908.00672/) (2019)
- [Bridging Composite and Real: Towards End-to-end Deep Image Matting](https://mlsanity.com/paper/2010.16188/) (2020)
- [Mask Guided Matting via Progressive Refinement Network](https://mlsanity.com/paper/2012.06722/) (2020)
- [Context-Aware Image Matting for Simultaneous Foreground and Alpha Estimation](https://mlsanity.com/paper/1909.09725/) (2019)
- [Natural Image Matting via Guided Contextual Attention](https://mlsanity.com/paper/2001.04069/) (2020)
- [Semantic Human Matting](https://mlsanity.com/paper/1809.01354/) (2018)
- [Privacy-Preserving Portrait Matting](https://mlsanity.com/paper/2104.14222/) (2021)
- [Information-Flow Matting](https://mlsanity.com/paper/1707.05055/) (2017)
- [Deep Automatic Natural Image Matting](https://mlsanity.com/paper/2107.07235/) (2021)
- [ViTMatte: Boosting Image Matting with Pretrained Plain Vision Transformers](https://mlsanity.com/paper/2305.15272/) (2023)
- [Boosting Semantic Human Matting with Coarse Annotations](https://mlsanity.com/paper/2004.04955/) (2020)
- [MatteFormer: Transformer-Based Image Matting via Prior-Tokens](https://mlsanity.com/paper/2203.15662/) (2022)
- [Rethinking Portrait Matting with Privacy Preserving](https://mlsanity.com/paper/2203.16828/) (2022)
- [Improved Image Matting via Real-time User Clicks and Uncertainty Estimation](https://mlsanity.com/paper/2012.08323/) (2020)
- [Long-Range Feature Propagating for Natural Image Matting](https://mlsanity.com/paper/2109.12252/) (2021)

## Built on by (most cited first, 4 of 4)

- [MatAnyone: Stable Video Matting with Consistent Memory Propagation](https://mlsanity.com/paper/2501.14677/) (2025)
- [MatAnyone 2: Scaling Video Matting via a Learned Quality Evaluator](https://mlsanity.com/paper/2512.11782/) (2025)
- [Capture Stage Matting: Challenges, Approaches, and Solutions for Offline and Real-Time Processing](https://mlsanity.com/paper/2507.07623/) (2025)
- [Self-supervised Automatic Matting](https://mlsanity.com/paper/2607.10395/) (2026)

## Abstract (arXiv)

The labelling difficulty has been a longstanding problem in deep image matting. To escape from fine labels, this work explores using rough annotations such as trimaps coarsely indicating the foreground/background as supervision. We present that the cooperation between learned semantics from indicated known regions and proper assumed matting rules can help infer alpha values at transition areas. Inspired by the nonlocal principle in traditional image matting, we build a directional distance consistency loss (DDC loss) at each pixel neighborhood to constrain the alpha values conditioned on the input image. DDC loss forces the distance of similar pairs on the alpha matte and on its corresponding image to be consistent. In this way, the alpha values can be propagated from learned known regions to unknown transition areas. With only images and trimaps, a matting model can be trained under the supervision of a known loss and the proposed DDC loss. Experiments on AM-2K and P3M-10K dataset show that our paradigm achieves comparable performance with the fine-label-supervised baseline, while sometimes offers even more satisfying results than human-labelled ground truth. Code is available at urlhttps://github.com/poppuppy/alpha-free-matting.