Neural implicit functions have achieved impressive results for reconstructing 3D shapes from single images. However, the image features for describing 3D point samplings of implicit functions are less effective when significant variations of occlusions, views, and appearances exist from the image. To better encode image features, we study a geometry-aware convolutional kernel to leverage geometric relationships of point samplings by the proposed \emph{spatial pattern}, i.e., a structured point set. Specifically, the kernel operates at 2D projections of 3D points from the spatial pattern. Supported by the spatial pattern, the 2D kernel encodes geometric information that is crucial for 3D reconstruction tasks, while traditional ones mainly consider appearance information. Furthermore, to enable the network to discover more adaptive spatial patterns for further capturing non-local contextual information, the kernel is devised to be deformable manipulated by a spatial pattern generator. Experimental results on both synthetic and real datasets demonstrate the superiority of the proposed method. Pre-trained models, codes, and data are available at https://github.com/yixin26/SVR-SP.
翻译:在从单个图像重建 3D 形状方面,内隐功能取得了令人印象深刻的结果。然而,如果图像中存在3D点的隐含功能样本的显著差异,描述隐含功能的3D点抽样的图像特征就不太有效。为了更好地对图像特征进行编码,我们研究几何-能进化的进化内核,以利用拟议的 emph{spatial 模式} (即一个结构化的点) 的点取样的几何关系。具体地说,内核在2D 预测空间模式3D 点时运行。在空间模式的支持下,2D 内核为对3D 重建任务至关重要的几何学信息编码提供了支持,而传统的内核内核则主要考虑外观信息。此外,为了使网络能够发现更适应性的空间模式,以进一步获取非本地的背景资料,内核内核将设计成由空间模式生成器操纵。合成和真实数据集的实验结果显示了拟议方法的优越性。在空间模式、代码和数据上经过预先训练的模型、代码和数据,可在 https://Syix26 。