In orthodontic treatment, a full tooth model consisting of both the crown and root is indispensable in making the treatment plan. However, acquiring tooth root information to obtain the full tooth model from CBCT images is sometimes restricted due to the massive radiation of CBCT scanning. Thus, reconstructing the full tooth shape from the ready-to-use input, e.g., the partial intra-oral scan and the 2D panoramic image, is an applicable and valuable solution. In this paper, we propose a neural network, called ToothInpaintor, that takes as input a partial 3D dental model and a 2D panoramic image and reconstructs the full tooth model with high-quality root(s). Technically, we utilize the implicit representation for both the 3D and 2D inputs, and learn a latent space of the full tooth shapes. At test time, given an input, we successfully project it to the learned latent space via neural optimization to obtain the full tooth model conditioned on the input. To help find the robust projection, a novel adversarial learning module is exploited in our pipeline. We extensively evaluate our method on a dataset collected from real-world clinics. The evaluation, comparison, and comprehensive ablation studies demonstrate that our approach produces accurate complete tooth models robustly and outperforms the state-of-the-art methods.
We present a novel method, called NeuralUDF, for reconstructing surfaces with arbitrary topologies from 2D images via volume rendering. Recent advances in neural rendering based reconstruction have achieved compelling results. However, these methods are limited to objects with closed surfaces since they adopt Signed Distance Function (SDF) as surface representation which requires the target shape to be divided into inside and outside. In this paper, we propose to represent surfaces as the Unsigned Distance Function (UDF) and develop a new volume rendering scheme to learn the neural UDF representation. Specifically, a new density function that correlates the property of UDF with the volume rendering scheme is introduced for robust optimization of the UDF fields. Experiments on the DTU and DeepFashion3D datasets show that our method not only enables high-quality reconstruction of non-closed shapes with complex typologies, but also achieves comparable performance to the SDF based methods on the reconstruction of closed surfaces.
In this paper, we introduce a set of effective TOken REduction (TORE) strategies for Transformer-based Human Mesh Recovery from monocular images. Current SOTA performance is achieved by Transformer-based structures. However, they suffer from high model complexity and computation cost caused by redundant tokens. We propose token reduction strategies based on two important aspects, i.e., the 3D geometry structure and 2D image feature, where we hierarchically recover the mesh geometry with priors from body structure and conduct token clustering to pass fewer but more discriminative image feature tokens to the Transformer. As a result, our method vastly reduces the number of tokens involved in high-complexity interactions in the Transformer, achieving competitive accuracy of shape recovery at a significantly reduced computational cost. We conduct extensive experiments across a wide range of benchmarks to validate the proposed method and further demonstrate the generalizability of our method on hand mesh recovery. Our code will be publicly available once the paper is published.
3D Morphable models of the human body capture variations among subjects and are useful in reconstruction and editing applications. Current dental models use an explicit mesh scene representation and model only the teeth, ignoring the gum. In this work, we present the first parametric 3D morphable dental model for both teeth and gum. Our model uses an implicit scene representation and is learned from rigidly aligned scans. It is based on a component-wise representation for each tooth and the gum, together with a learnable latent code for each of such components. It also learns a template shape thus enabling several applications such as segmentation, interpolation, and tooth replacement. Our reconstruction quality is on par with the most advanced global implicit representations while enabling novel applications. Project page: https://vcai.mpi-inf.mpg.de/projects/DMM/
Sherds, as the most common artifacts uncovered during archaeological excavations, carry rich information about past human societies so need to be accurately reconstructed and recorded digitally for analysis and preservation. Often hundreds of fragments are uncovered in a day at an archaeological excavation site, far beyond the scanning capacity of existing imaging systems. Hence, there is high demand for a desirable image acquisition system capable of imaging hundreds of fragments per day. In response to this demand, we developed a new system, dubbed FIRES, for Fast Imaging and 3D REconstruction of Sherds. The FIRES system consists of two main components. The first is an optimally designed fast image acquisition device capable of capturing over 700 sherds per day (in 8 working hours) in actual tests at an excavation site, which is one order-of-magnitude faster than existing systems. The second component is an automatic pipeline for 3D reconstruction of the sherds from the images captured by the imaging acquisition system, achieving reconstruction accuracy of 0.16 milimeters. The pipeline includes a novel batch matching algorithm that matches partial 3D scans of the front and back sides of the sherds and a new ICP-type method that registers the front and back sides sharing very narrow overlapping regions. Extensive validation in labs and testing in excavation sites demonstrated that our FIRES system provides the first fast, accurate, portal, and cost-effective solution for the task of imaging and 3D reconstruction of sherds in archaeological excavations.
We investigate transductive zero-shot point cloud semantic segmentation in this paper, where unseen class labels are unavailable during training. Actually, the 3D geometric elements are essential cues to reason the 3D object type. If two categories share similar geometric primitives, they also have similar semantic representations. Based on this consideration, we propose a novel framework to learn the geometric primitives shared in seen and unseen categories' objects, where the learned geometric primitives are served for transferring knowledge from seen to unseen categories. Specifically, a group of learnable prototypes automatically encode geometric primitives via back-propagation. Then, the point visual representation is formulated as the similarity vector of its feature to the prototypes, which implies semantic cues for both seen and unseen categories. Besides, considering a 3D object composed of multiple geometric primitives, we formulate the semantic representation as a mixture-distributed embedding for the fine-grained match of visual representation. In the end, to effectively learn the geometric primitives and alleviate the misclassification issue, we propose a novel unknown-aware infoNCE loss to align the visual and semantic representation. As a result, guided by semantic representations, the network recognizes the novel object represented with geometric primitives. Extensive experiments show that our method significantly outperforms other state-of-the-art methods in the harmonic mean-intersection-over-union (hIoU), with the improvement of 17.8%, 30.4% and 9.2% on S3DIS, ScanNet and SemanticKITTI datasets, respectively. Codes will be released.
Understanding dynamic hand motions and actions from egocentric RGB videos is a fundamental yet challenging task due to self-occlusion and ambiguity. To address occlusion and ambiguity, we develop a transformer-based framework to exploit temporal information for robust estimation. Noticing the different temporal granularity of and the semantic correlation between hand pose estimation and action recognition, we build a network hierarchy with two cascaded transformer encoders, where the first one exploits the short-term temporal cue for hand pose estimation, and the latter aggregates per-frame pose and object information over a longer time span to recognize the action. Our approach achieves competitive results on two first-person hand action benchmarks, namely FPHA and H2O. Extensive ablation studies verify our design choices. We will open-source code and data to facilitate future research.
Estimating the pose of a moving camera from monocular video is a challenging problem, especially due to the presence of moving objects in dynamic environments, where the performance of existing camera pose estimation methods are susceptible to pixels that are not geometrically consistent. To tackle this challenge, we present a robust dense indirect structure-from-motion method for videos that is based on dense correspondence initialized from pairwise optical flow. Our key idea is to optimize long-range video correspondence as dense point trajectories and use it to learn robust estimation of motion segmentation. A novel neural network architecture is proposed for processing irregular point trajectory data. Camera poses are then estimated and optimized with global bundle adjustment over the portion of long-range point trajectories that are classified as static. Experiments on MPI Sintel dataset show that our system produces significantly more accurate camera trajectories compared to existing state-of-the-art methods. In addition, our method is able to retain reasonable accuracy of camera poses on fully static scenes, which consistently outperforms strong state-of-the-art dense correspondence based methods with end-to-end deep learning, demonstrating the potential of dense indirect methods based on optical flow and point trajectories. As the point trajectory representation is general, we further present results and comparisons on in-the-wild monocular videos with complex motion of dynamic objects. Code is available at https://github.com/bytedance/particle-sfm.
This paper presents a Progressively-connected Light Field network (ProLiF), for the novel view synthesis of complex forward-facing scenes. ProLiF encodes a 4D light field, which allows rendering a large batch of rays in one training step for image- or patch-level losses. Directly learning a neural light field from images has difficulty in rendering multi-view consistent images due to its unawareness of the underlying 3D geometry. To address this problem, we propose a progressive training scheme and regularization losses to infer the underlying geometry during training, both of which enforce the multi-view consistency and thus greatly improves the rendering quality. Experiments demonstrate that our method is able to achieve significantly better rendering quality than the vanilla neural light fields and comparable results to NeRF-like rendering methods on the challenging LLFF dataset and Shiny Object dataset. Moreover, we demonstrate better compatibility with LPIPS loss to achieve robustness to varying light conditions and CLIP loss to control the rendering style of the scene. Project page: https://totoro97.github.io/projects/prolif.