We propose a new approach to 3D reconstruction from sequences of images acquired by monocular endoscopes. It is based on two key insights. First, endoluminal cavities are watertight, a property naturally enforced by modeling them in terms of a signed distance function. Second, the scene illumination is variable. It comes from the endoscope's light sources and decays with the inverse of the squared distance to the surface. To exploit these insights, we build on NeuS, a neural implicit surface reconstruction technique with an outstanding capability to learn appearance and a SDF surface model from multiple views, but currently limited to scenes with static illumination. To remove this limitation and exploit the relation between pixel brightness and depth, we modify the NeuS architecture to explicitly account for it and introduce a calibrated photometric model of the endoscope's camera and light source. Our method is the first one to produce watertight reconstructions of whole colon sections. We demonstrate excellent accuracy on phantom imagery. Remarkably, the watertight prior combined with illumination decline, allows to complete the reconstruction of unseen portions of the surface with acceptable accuracy, paving the way to automatic quality assessment of cancer screening explorations, measuring the global percentage of observed mucosa.
Single-view depth estimation can be remarkably effective if there is enough ground-truth depth data for supervised training. However, there are scenarios, especially in medicine in the case of endoscopies, where such data cannot be obtained. In such cases, multi-view self-supervision and synthetic-to-real transfer serve as alternative approaches, however, with a considerable performance reduction in comparison to supervised case. Instead, we propose a single-view self-supervised method that achieves a performance similar to the supervised case. In some medical devices, such as endoscopes, the camera and light sources are co-located at a small distance from the target surfaces. Thus, we can exploit that, for any given albedo and surface orientation, pixel brightness is inversely proportional to the square of the distance to the surface, providing a strong single-view self-supervisory signal. In our experiments, our self-supervised models deliver accuracies comparable to those of fully supervised ones, while being applicable without depth ground-truth data.
Objects interact with each other in various ways, including containment, contact, or maintaining fixed distances. Ensuring these topological interactions is crucial for accurate modeling in many scenarios. In this paper, we propose a novel method to refine 3D object representations, ensuring that their surfaces adhere to a topological prior. Our key observation is that the object interaction can be observed via a stochastic approximation method: the statistic of signed distances between a large number of random points to the object surfaces reflect the interaction between them. Thus, the object interaction can be indirectly manipulated by using choosing a set of points as anchors to refine the object surfaces. In particular, we show that our method can be used to enforce two objects to have a specific contact ratio while having no surface intersection. The conducted experiments show that our proposed method enables accurate 3D reconstruction of human hearts, ensuring proper topological connectivity between components. Further, we show that our proposed method can be used to simulate various ways a hand can interact with an arbitrary object.
Many approaches to draping individual garments on human body models are realistic, fast, and yield outputs that are differentiable with respect to the body shape on which they are draped. However, none of them can handle multi-layered clothing, which is prevalent in everyday dress. In this paper, we introduce a parametric garment representation model that can. As in models used by clothing designers, each garment consists of individual 2D panels. Their 2D shape is defined by a Signed Distance Function and 3D shape by a 2D to 3D mapping. The 2D parameterization enables easy detection of potential collisions and the 3D parameterization handles complex shapes effectively. We show that this combination is faster and yields higher quality reconstructions than purely implicit surface representations, and makes the recovery of layered garments from images possible thanks to its differentiability. Furthermore, it supports rapid editing of garment shapes and texture by modifying individual 2D panels.
We propose two deep learning models that fully automate shape parameterization for aerodynamic shape optimization. Both models are optimized to parameterize via deep geometric learning to embed human prior knowledge into learned geometric patterns, eliminating the need for further handcrafting. The Latent Space Model (LSM) learns a low-dimensional latent representation of an object from a dataset of various geometries, while the Direct Mapping Model (DMM) builds parameterization on the fly using only one geometry of interest. We also devise a novel regularization loss that efficiently integrates volumetric mesh deformation into the parameterization model. The models directly manipulate the high-dimensional mesh data by moving vertices. LSM and DMM are fully differentiable, enabling gradient-based, end-to-end pipeline design and plug-and-play deployment of surrogate models or adjoint solvers. We perform shape optimization experiments on 2D airfoils and discuss the applicable scenarios for the two models.
Diffusion models generating images conditionally on text, such as Dall-E 2 and Stable Diffusion, have recently made a splash far beyond the computer vision community. Here, we tackle the related problem of generating point clouds, both unconditionally, and conditionally with images. For the latter, we introduce a novel geometrically-motivated conditioning scheme based on projecting sparse image features into the point cloud and attaching them to each individual point, at every step in the denoising process. This approach improves geometric consistency and yields greater fidelity than current methods relying on unstructured, global latent codes. Additionally, we show how to apply recent continuous-time diffusion schemes. Our method performs on par or above the state of art on conditional and unconditional experiments on synthetic data, while being faster, lighter, and delivering tractable likelihoods. We show it can also scale to diverse indoors scenes.
Vision transformers have emerged as powerful tools for many computer vision tasks. It has been shown that their features and class tokens can be used for salient object segmentation. However, the properties of segmentation transformers remain largely unstudied. In this work we conduct an in-depth study of the spatial attentions of different backbone layers of semantic segmentation transformers and uncover interesting properties. The spatial attentions of a patch intersecting with an object tend to concentrate within the object, whereas the attentions of larger, more uniform image areas rather follow a diffusive behavior. In other words, vision transformers trained to segment a fixed set of object classes generalize to objects well beyond this set. We exploit this by extracting heatmaps that can be used to segment unknown objects within diverse backgrounds, such as obstacles in traffic scenes. Our method is training-free and its computational overhead negligible. We use off-the-shelf transformers trained for street-scene segmentation to process other scene types.
Recent approaches to drape garments quickly over arbitrary human bodies leverage self-supervision to eliminate the need for large training sets. However, they are designed to train one network per clothing item, which severely limits their generalization abilities. In our work, we rely on self-supervision to train a single network to drape multiple garments. This is achieved by predicting a 3D deformation field conditioned on the latent codes of a generative network, which models garments as unsigned distance fields. Our pipeline can generate and drape previously unseen garments of any topology, whose shape can be edited by manipulating their latent codes. Being fully differentiable, our formulation makes it possible to recover accurate 3D models of garments from partial observations -- images or 3D scans -- via gradient descent. Our code will be made publicly available.
Given enough annotated training data, 3D human pose estimation models can achieve high accuracy. However, annotations are not always available, especially for people performing unusual activities. In this paper, we propose an algorithm that learns to detect 3D keypoints on human bodies from multiple-views without any supervision other than the constraints multiple-view geometry provides. To ensure that the estimated 3D keypoints are meaningful, they are re-projected to each view to estimate the person's mask that the model itself has initially estimated. Our approach outperforms other state-of-the-art unsupervised 3D human pose estimation methods on the Human3.6M and MPI-INF-3DHP benchmark datasets.
Multi-task learning is central to many real-world applications. Unfortunately, obtaining labelled data for all tasks is time-consuming, challenging, and expensive. Active Learning (AL) can be used to reduce this burden. Existing techniques typically involve picking images to be annotated and providing annotations for all tasks. In this paper, we show that it is more effective to select not only the images to be annotated but also a subset of tasks for which to provide annotations at each AL iteration. Furthermore, the annotations that are provided can be used to guess pseudo-labels for the tasks that remain unannotated. We demonstrate the effectiveness of our approach on several popular multi-task datasets.