Weakly- and Semi-Supervised Learning of a Deep Convolutional Network for Semantic Image Segmentation
George Papandreou Google, Inc. The first two authors contributed equally to this work. Liang-Chieh Chen∗ UCLA Kevin Murphy Google, Inc. Alan L. Yuille UCLA
Abstract
Deep convolutional neural networks (DCNNs) trained on a large number of images with strong pixel-level annotations have recently significantly pushed the state-of-art in semantic image segmentation. We study the more challenging problem of learning DCNNs for semantic image segmentation from either (1) weakly annotated training data such as bounding boxes or image-level labels or (2) a combination of few strongly labeled and many weakly labeled images, sourced from one or multiple datasets. We develop Expectation-Maximization (EM) methods for semantic image segmentation model training under these weakly supervised and semi-supervised settings. Extensive experimental evaluation shows that the proposed techniques can learn models delivering competitive results on the challenging PASCAL VOC 2012 image segmentation benchmark, while requiring significantly less annotation effort. We share source code implementing the proposed system at https://bitbucket.org/deeplab/deeplab-public.
中文速览
语义图像分割（semantic image segmentation）模型通常需要昂贵的逐像素标注数据才能训练，而现实中大量图像只有边界框或图像级别标签这类廉价弱标注。研究者基于 DeepLab-CRF 框架，提出了一套期望最大化（Expectation-Maximization，EM）算法，在训练过程中交替推断每张图的潜在像素分割和更新网络参数，从而充分利用弱标注信息。实验表明，仅用图像级别标签可达到 39.6% 的 IOU，仅用边界框可达 62.2%，而将少量像素级标注与大量弱标注混合使用时，性能可接近完全监督水平（68.5% vs. 70.3%），跨数据集联合训练进一步将结果推高至 73.9%。这项工作证明了在标注预算有限的现实场景中，通过巧妙的半监督和弱监督算法设计，深度分割模型同样可以达到接近全监督的效果，极大降低了高质量分割系统的数据采集成本。
原文 arXiv:1502.02734;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1502.02734v3