ID 原文 译文
38846 本模型在自行采集的中国手语数据集上进行测试，得到了高达0.943的准确率。 This model was tested on a Chinese sign language data set collected by ourselves, and obtained a high accuracy rate 0.943.
38847 实例分割，又名同时检测和分割(simultaneous detection and segmentation)，需要标注像素级别的实例掩膜用于训练。 Instance segmentation, also known as simultaneous detection and segmentation(SDS) requires pixel-level instance masks during training.
38848 然而，这种标注工作需要非常细致的人力劳动，费时费力。 It makes data preparation for mask annotation a labor-intensive task.
38849 本论文提出只使用每个目标实例的单点标注，使得标注成本大大降低。 In this paper, we only use a one-point label for each object instance, which is easy to draw.
38850 本文提出的模型包括两个模块:基于外观信息和相邻包围框投票的框校验模块，以及基于推断掩膜的上下文信息的区块校验模块。 Our training consists of box verification, which is based on appearance and voting of neighboring boxes, and segment verification on context information of proposal masks.
38851 这种设计保留了像素级别的实例信息，有助于抑制单纯图像分割模型训练过程中的误差累积。 This structure preserves pixel-wise instance information and helps prevent error accumulation compared with trivially training a single segmentation model iteratively.
38852 本文使用弱监督和半监督训练的实验来验证本工作的有效性，比现有方法取得更高的实例分割性能。 We conduct weakly-and semi-supervised experiments to manifest that this design is effective. Our approach surpasses the state-of-the-art methods.
38853 单幅图像的深度估计是场景几何理解过程中的一个重要步骤，但由于尺度模糊，也被计算机视觉领域普遍认为是一个典型的不适定问题。 Monocular depth estimation is an important but ill-posed procedure in the process of scene geometry understanding.
38854 近年来，尽管监督学习方法在单目深度估计中取得了基本令人满意的效果，但需要对数据集进行大量真实深度值的标记，这是一项成本较高的工作。 Though recent supervised learning methods have achieved promising results for monocular depth estimation, they require vast amounts of ground truth depth data which is a costly task.
38855 此外，由于物体的运动、遮挡、光照等常见问题，单目深度估计的表现并不尽如人意，尤其是在物体边缘和弱纹理区域。 Besides, previous works suffer from well-known problems such as moving objects, occlusions and lighting, which result in unsatisfactory performance, particularly in object edges and low-texture regions.