Getting to 99% Accuracy in Interactive Segmentation
This work addresses the problem of improving precision in interactive object cutout tools for image editing users, representing a strong specific gain rather than a foundational advancement.
The paper tackled the plateau in accuracy of deep-learning-based interactive segmentation tools after initial rough selections, proposing a novel architecture and training scheme tailored to user workflow and a synthetic dataset for complex boundaries, achieving state-of-the-art performance with 99% accuracy.
Interactive object cutout tools are the cornerstone of the image editing workflow. Recent deep-learning based interactive segmentation algorithms have made significant progress in handling complex images and rough binary selections can typically be obtained with just a few clicks. Yet, deep learning techniques tend to plateau once this rough selection has been reached. In this work, we interpret this plateau as the inability of current algorithms to sufficiently leverage each user interaction and also as the limitations of current training/testing datasets. We propose a novel interactive architecture and a novel training scheme that are both tailored to better exploit the user workflow. We also show that significant improvements can be further gained by introducing a synthetic training dataset that is specifically designed for complex object boundaries. Comprehensive experiments support our approach, and our network achieves state of the art performance.