The paper introduces Inf-Net, a novel deep network designed to automatically identify infected regions in chest CT slices for COVID-19 lung infection segmentation. The network employs a parallel partial decoder (PPD) to aggregate high-level features and generate a global map, followed by reverse attention (RA) and edge-attention (EA) modules to model boundaries and enhance representations. To address the challenge of limited labeled data, a semi-supervised segmentation framework is proposed, which uses a randomly selected propagation strategy to leverage unlabeled data. Extensive experiments on the COVID-SemiSeg dataset and real CT volumes demonstrate that Inf-Net outperforms most cutting-edge segmentation models, advancing the state-of-the-art performance. The paper also discusses related work in chest CT segmentation, semi-supervised learning, and AI for COVID-19, and provides implementation details and ablation studies to validate the effectiveness of each component.The paper introduces Inf-Net, a novel deep network designed to automatically identify infected regions in chest CT slices for COVID-19 lung infection segmentation. The network employs a parallel partial decoder (PPD) to aggregate high-level features and generate a global map, followed by reverse attention (RA) and edge-attention (EA) modules to model boundaries and enhance representations. To address the challenge of limited labeled data, a semi-supervised segmentation framework is proposed, which uses a randomly selected propagation strategy to leverage unlabeled data. Extensive experiments on the COVID-SemiSeg dataset and real CT volumes demonstrate that Inf-Net outperforms most cutting-edge segmentation models, advancing the state-of-the-art performance. The paper also discusses related work in chest CT segmentation, semi-supervised learning, and AI for COVID-19, and provides implementation details and ablation studies to validate the effectiveness of each component.