Semi-Supervised Wide-Angle Portraits Correction by Multi-Scale Transformer
This addresses a domain-specific problem in computer vision for portrait photography, offering a more efficient solution by reducing reliance on expensive manual annotations, though it is incremental as it builds on existing deep learning approaches.
The paper tackles the problem of correcting skew and distortion in wide-angle portraits, which is costly due to the need for manual ground-truth labels, by proposing a semi-supervised network that uses both labeled and unlabeled data with novel consistency mechanisms and a multi-scale transformer architecture, achieving superior performance over state-of-the-art methods.
We propose a semi-supervised network for wide-angle portraits correction. Wide-angle images often suffer from skew and distortion affected by perspective distortion, especially noticeable at the face regions. Previous deep learning based approaches need the ground-truth correction flow maps for training guidance. However, such labels are expensive, which can only be obtained manually. In this work, we design a semi-supervised scheme and build a high-quality unlabeled dataset with rich scenarios, allowing us to simultaneously use labeled and unlabeled data to improve performance. Specifically, our semi-supervised scheme takes advantage of the consistency mechanism, with several novel components such as direction and range consistency (DRC) and regression consistency (RC). Furthermore, different from the existing methods, we propose the Multi-Scale Swin-Unet (MS-Unet) based on the multi-scale swin transformer block (MSTB), which can simultaneously learn short-distance and long-distance information to avoid artifacts. Extensive experiments demonstrate that the proposed method is superior to the state-of-the-art methods and other representative baselines. The source code and dataset are available at: https://github.com/megvii-research/Portraits_Correction.