The paper introduces a novel bilateral reference framework (BiRefNet) for high-resolution dichotomous image segmentation (DIS). BiRefNet consists of two main components: the localization module (LM) and the reconstruction module (RM) with a proposed bilateral reference (BiRef). The LM aids in object localization using global semantic information, while the RM uses BiRef for reconstruction, where hierarchical patches of images serve as source references and gradient maps as target references. The paper also introduces auxiliary gradient supervision to enhance focus on regions with finer details. Practical training strategies tailored for DIS are outlined to improve map quality and training efficiency. Extensive experiments on four tasks—DIS, high-resolution salient object detection (HRSOD), concealed object detection (COD), and salient object detection (SOD)—show that BiRefNet outperforms task-specific cutting-edge methods across all benchmarks. The codes are publicly available at <https://github.com/ZhengPeng7/BiRefNet>.The paper introduces a novel bilateral reference framework (BiRefNet) for high-resolution dichotomous image segmentation (DIS). BiRefNet consists of two main components: the localization module (LM) and the reconstruction module (RM) with a proposed bilateral reference (BiRef). The LM aids in object localization using global semantic information, while the RM uses BiRef for reconstruction, where hierarchical patches of images serve as source references and gradient maps as target references. The paper also introduces auxiliary gradient supervision to enhance focus on regions with finer details. Practical training strategies tailored for DIS are outlined to improve map quality and training efficiency. Extensive experiments on four tasks—DIS, high-resolution salient object detection (HRSOD), concealed object detection (COD), and salient object detection (SOD)—show that BiRefNet outperforms task-specific cutting-edge methods across all benchmarks. The codes are publicly available at <https://github.com/ZhengPeng7/BiRefNet>.