No file available
To read the file of this research,
you can request a copy directly from the authors.
Reconstructing Three-Dimensional Models of Interacting Humans
Preprints and early-stage research may not have been peer reviewed yet.
Abstract
Understanding 3d human interactions is fundamental for fine-grained scene analysis and behavioural modeling. However, most of the existing models predict incorrect, lifeless 3d estimates, that miss the subtle human contact aspects--the essence of the event--and are of little use for detailed behavioral understanding. This paper addresses such issues with several contributions: (1) we introduce models for interaction signature estimation (ISP) encompassing contact detection, segmentation, and 3d contact signature prediction; (2) we show how such components can be leveraged to ensure contact consistency during 3d reconstruction; (3) we construct several large datasets for learning and evaluating 3d contact prediction and reconstruction methods; specifically, we introduce CHI3D, a lab-based accurate 3d motion capture dataset with 631 sequences containing 2,525 contact events, 728,664 ground truth 3d poses, as well as FlickrCI3D, a dataset of 11,216 images, with 14,081 processed pairs of people, and 81,233 facet-level surface correspondences. Finally, (4) we propose methodology for recovering the ground-truth pose and shape of interacting people in a controlled setup and (5) annotate all 3d interaction motions in CHI3D with textual descriptions. Motion data in multiple formats (GHUM and SMPLX parameters, Human3.6m 3d joints) is made available for research purposes at \url{https://ci3d.imar.ro}, together with an evaluation server and a public benchmark.
ResearchGate has not been able to resolve any citations for this publication.
To understand and analyze human behavior, we need to capture humans moving in, and interacting with, the world. Most existing methods perform 3D human pose estimation without explicitly considering the scene. We observe however that the world constrains the body and vice-versa. To motivate this, we show that current 3D human pose estimation methods produce results that are not consistent with the 3D scene. Our key contribution is to exploit static 3D scene structure to better estimate human pose from monocular images. The method enforces Proximal Relationships with Object eXclusion and is called PROX. To test this, we collect a new dataset composed of 12 different 3D scenes and RGB sequences of 20 subjects moving in and interacting with the scenes. We represent human pose using the 3D human body model SMPL-X and extend SMPLify-X to estimate body pose using scene constraints. We make use of the 3D scene information by formulating two main constraints. The interpenetration constraint penalizes intersection between the body model and the surrounding 3D scene. The contact constraint encourages specific parts of the body to be in contact with scene surfaces if they are close enough in distance and orientation. For quantitative evaluation we capture a separate dataset with 180 RGB frames in which the ground-truth body pose is estimated using a motion-capture system. We show quantitatively that introducing scene constraints significantly reduces 3D joint error and vertex error. Our code and data are available for research at https://prox.is.tue.mpg.de.
We present a novel method for real-time pose and shape reconstruction of two strongly interacting hands. Our approach is the first two-hand tracking solution that combines an extensive list of favorable properties, namely it is marker-less, uses a single consumer-level depth camera, runs in real time, handles inter- and intra-hand collisions, and automatically adjusts to the user's hand shape. In order to achieve this, we embed a recent parametric hand pose and shape model and a dense correspondence predictor based on a deep neural network into a suitable energy minimization framework. For training the correspondence prediction network, we synthesize a two-hand dataset based on physical simulations that includes both hand pose and shape annotations while at the same time avoiding inter-hand penetrations. To achieve real-time rates, we phrase the model fitting in terms of a nonlinear least-squares problem so that the energy can be optimized based on a highly efficient GPU-based Gauss-Newton optimizer. We show state-of-the-art results in scenes that exceed the complexity level demonstrated by previous work, including tight two-hand grasps, significant inter-hand occlusions, and gesture interaction.
Action recognition and human pose estimation are closely related but both problems are generally handled as distinct tasks in the literature. In this work, we propose a multitask framework for jointly 2D and 3D pose estimation from still images and human action recognition from video sequences. We show that a single architecture can be used to solve the two problems in an efficient way and still achieves state-of-the-art results. Additionally, we demonstrate that optimization from end-to-end leads to significantly higher accuracy than separated learning. The proposed architecture can be trained with data from different categories simultaneously in a seamlessly way. The reported results on four datasets (MPII, Human3.6M, Penn Action and NTU) demonstrate the effectiveness of our method on the targeted tasks.
We consider the problem of estimating realistic contact forces during manipulation, backed with ground-truth measurements, using vision alone. Interaction forces are usually measured by mounting force transducers onto the manipulated objects or the hands. Those are costly, cumbersome, and alter the objects' physical properties and their perception by the human sense of touch. Our work establishes that interaction forces can be estimated in a cost-effective, reliable, non-intrusive way using vision. This is a complex and challenging problem. Indeed, in multi-contact, a given motion can generally be caused by an infinity of possible force distributions. To alleviate the limitations of traditional models based on inverse optimization, we collect and release the first large-scale dataset on manipulation kinodynamics as 3.2 hours of synchronized force and motion measurements under 193 object-grasp configurations. We learn a mapping between high-level kinematic features based on the equations of motion and the underlying manipulation forces using recurrent neural networks (RNN). The RNN predictions are consistently refined using physics-based optimization through second-order cone programming (SOCP). We show that our method can successfully capture interaction forces compatible with both the observations and the way humans intuitively manipulate objects, using a single RGB-D camera.
During occupational therapy for children with autism, it is often necessary to elicit and maintain engagement for the children to benefit from the session. Recently, social robots have been used for this; however, existing robots lack the ability to autonomously recognize the children’s level of engagement, which is necessary when choosing an optimal interaction strategy. Progress in automated engagement reading has been impeded in part due to a lack of studies on child-robot engagement in autism therapy. While it is well known that there are large individual differences in autism, little is known about how these vary across cultures. To this end, we analyzed the engagement of children (age 3–13) from two different cultural backgrounds: Asia (Japan, n = 17) and Eastern Europe (Serbia, n = 19). The children participated in a 25 min therapy session during which we studied the relationship between the children’s behavioral engagement (task-driven) and different facets of affective engagement (valence and arousal). Although our results indicate that there are statistically significant differences in engagement displays in the two groups, it is difficult to make any causal claims about these differences due to the large variation in age and behavioral severity of the children in the study. However, our exploratory analysis reveals important associations between target engagement and perceived levels of valence and arousal, indicating that these can be used as a proxy for the children’s engagement during the therapy. We provide suggestions on how this can be leveraged to optimize social robots for autism therapy, while taking into account cultural differences.
Following the success of deep convolutional networks, state-of-the-art methods for 3d human pose estimation have focused on deep end-to-end systems that predict 3d joint locations given raw image pixels. Despite their excellent performance, it is often not easy to understand whether their remaining error stems from a limited 2d pose (visual) understanding, or from a failure to map 2d poses into 3-dimensional positions. With the goal of understanding these sources of error, we set out to build a system that given 2d joint locations predicts 3d positions. Much to our surprise, we have found that, with current technology, "lifting" ground truth 2d joint locations to 3d space is a task that can be solved with a remarkably low error rate: a relatively simple deep feed-forward network outperforms the best reported result by about 30% on Human3.6M, the largest publicly available 3d pose estimation benchmark. Furthermore, training our system on the output of an off-the-shelf state-of-the-art 2d detector (i.e., using images as input) yields results on par with the state of the art -- this includes an array of systems that have been trained end-to-end specifically for this task. Our results indicate that a large portion of the error of state-of-the-art deep 3d pose estimation systems stems from their visual analysis, and suggests directions to further advance the state of the art in 3d human pose estimation.
We propose a deep multitask architecture for \emph{fully automatic 2d and 3d human sensing} (DMHS), including \emph{recognition and reconstruction}, in \emph{monocular images}. The system computes the figure-ground segmentation, semantically identifies the human body parts at pixel level, and estimates the 2d and 3d pose of the person. The model supports the joint training of all components by means of multi-task losses where early processing stages recursively feed into advanced ones for increasingly complex calculations, accuracy and robustness. The design allows us to tie a complete training protocol, by taking advantage of multiple datasets that would otherwise restrictively cover only some of the model components: complex 2d image data with no body part labeling and without associated 3d ground truth, or complex 3d data with limited 2d background variability. In detailed experiments based on several challenging 2d and 3d datasets (LSP, HumanEva, Human3.6M), we evaluate the sub-structures of the model, the effect of various types of training data in the multitask loss, and demonstrate that state-of-the-art results can be achieved at all processing levels. We also show that in the wild our monocular RGB architecture is perceptually competitive to a state-of-the art (commercial) Kinect system based on RGB-D data.
Engagement is crucial to designing intelligent systems that can adapt to the characteristics of their users. This paper focuses on automatic analysis and classification of engagement based on humans’ and robot’s personality profiles in a triadic human-human-robot interaction setting. More explicitly, we present a study that involves two participants interacting with a humanoid robot, and investigate how participants’ personalities can be used together with the robot’s personality to predict the engagement state of each participant. The fully automatic system is firstly trained to predict the Big Five personality traits of each participant by extracting individual and interpersonal features from their nonverbal behavioural cues. Secondly, the output of the personality prediction system is used as an input to the engagement classification system. Thirdly, we focus on the concept of “group engagement”, which we define as the collective engagement of the participants with the robot, and analyse the impact of similar and dissimilar personalities on the engagement classification. Our experimental results show that (i) using the automatically predicted personality labels for engagement classification yields an F-measure on par with using the manually annotated personality labels, demonstrating the effectiveness of the automatic personality prediction module proposed; (ii) using the individual and interpersonal features without utilising personality information is not sufficient for engagement classification, instead incorporating the participants’ and robot’s personalities with individual/interpersonal features increases engagement classification performance; and (iii) the best classification performance is achieved when the participants and the robot are extroverted, while the worst results are obtained when all are introverted.
Studying early interaction is essential for understanding development and psychopathology. Automatic computational methods offer the possibility to analyse social signals and behaviours of several partners simultaneously and dynamically. Here, 20 dyads of mothers and their 13-36-month-old infants were videotaped during mother-infant interaction including 10 extremely high-risk and 10 low-risk dyads using two-dimensional (2D) and three-dimensional (3D) sensors. From 2D+3D data and 3D space reconstruction, we extracted individual parameters (quantity of movement and motion activity ratio for each partner) and dyadic parameters related to the dynamics of partners heads distance (contribution to heads distance), to the focus of mutual engagement (percentage of time spent face to face or oriented to the task) and to the dynamics of motion activity (synchrony ratio, overlap ratio, pause ratio). Features are compared with blind global rating of the interaction using the coding interactive behavior (CIB). We found that individual and dyadic parameters of 2D+3D motion features perfectly correlates with rated CIB maternal and dyadic composite scores. Support Vector Machine classification using all 2D-3D motion features classified 100% of the dyads in their group meaning that motion behaviours are sufficient to distinguish high-risk from low-risk dyads. The proposed method may present a promising, low-cost methodology that can uniquely use artificial technology to detect meaningful features of human interactions and may have several implications for studying dyadic behaviours in psychiatry. Combining both global rating scales and computerized methods may enable a continuum of time scale from a summary of entire interactions to second-by-second dynamics.
Human motion analysis in images and video, with its deeply inter-related 2D and 3D inference components, is a central computer vision problem. Yet, there are no studies that reveal how humans perceive other people in images and how accurate they are. In this paper we aim to unveil some of the processing—as well as the levels of accuracy—involved in the 3D perception of people from images by assessing the human performance. Moreover, we reveal the quantitative and qualitative differences between human and computer performance when presented with the same visual stimuli and show that metrics incorporating human perception can produce more meaningful results when integrated into automatic pose prediction algorithms. Our contributions are: (1) the construction of an experimental apparatus that relates perception and measurement, in particular the visual and kinematic performance with respect to 3D ground truth when the human subject is presented an image of a person in a given pose; (2) the creation of a dataset containing images, articulated 2D and 3D pose ground truth, as well as synchronized eye movement recordings of human subjects, shown a variety of human body configurations, both easy and difficult, as well as their ‘re-enacted’ 3D poses; (3) quantitative analysis revealing the human performance in 3D pose re-enactment tasks, the degree of stability in the visual fixation patterns of human subjects, and the way it correlates with different poses; (4) extensive analysis on the differences between human re-enactments and poses produced by an automatic system when presented with the same visual stimuli; (5) an approach to learning perceptual metrics that, when integrated into visual sensing systems, produces more stable and meaningful results.
Significance
Touch is a powerful tool for communicating positive emotions. However, it has remained unknown to what extent social touch would maintain and establish social bonds. We asked a total of 1,368 people from five countries to reveal, using an Internet-based topographical self-reporting tool, those parts of their body that they would allow relatives, friends, and strangers to touch. These body regions formed relationship-specific maps in which the total area was directly related to the strength of the emotional bond between the participant and the touching person. Cultural influences were minor. We suggest that these relation-specific bodily patterns of social touch constitute an important mechanism supporting the maintenance of human social bonds.
Hand motion capture is a popular research field, recently gaining more
attention due to the ubiquity of RGB-D sensors. However, even most recent
approaches focus on the case of a single isolated hand. In this work, we focus
on hands that interact with other hands or objects and present a framework that
successfully captures motion in such interaction scenarios for both rigid and
articulated objects. Our framework combines a generative model with
discriminatively trained salient points to achieve a low tracking error and
with collision detection and physics simulation to achieve physically plausible
estimates even in case of occlusions and missing visual data. Since all
components are unified in a single objective function which is almost
everywhere differentiable, it can be optimized with standard optimization
techniques. Our approach works for monocular RGB-D sequences as well as setups
with multiple synchronized RGB cameras. For a qualitative and quantitative
evaluation, we captured 29 sequences with a large variety of interactions and
up to 150 degrees of freedom.
Capturing the motion of two hands interacting with an object is a very challenging task due to the large number of degrees of freedom, self-occlusions, and similarity between the fingers, even in the case of multiple cameras observing the scene. In this paper we propose to use discriminatively learned salient points on the fingers and to estimate the finger-salient point associations simultaneously with the estimation of the hand pose. We introduce a differentiable objective function that also takes edges, optical flow and collisions into account. Our qualitative and quantitative evaluations show that the proposed approach achieves very accurate results for several challenging sequences containing hands and objects in action.
Human activity recognition has potential to impact a wide range of applications from surveillance to human computer interfaces to content based video retrieval. Recently, the rapid development of inexpensive depth sensors (e.g. Microsoft Kinect) provides adequate accuracy for real-time full-body human tracking for activity recognition applications. In this paper, we create a complex human activity dataset depicting two person interactions, including synchronized video, depth and motion capture data. Moreover, we use our dataset to evaluate various features typically used for indexing and retrieval of motion capture data, in the context of real-time detection of interaction activities via Support Vector Machines (SVMs). Experimentally, we find that the geometric relational features based on distance between all pairs of joints outperforms other feature choices. For whole sequence classification, we also explore techniques related to Multiple Instance Learning (MIL) in which the sequence is represented by a bag of body-pose features. We find that the MIL based classifier outperforms SVMs when the sequences extend temporally around the interaction of interest.
Benchmarking methods for 3d hand tracking is still an open problem due to the difficulty of acquiring ground truth data. We in-troduce a new dataset and benchmarking protocol that is insensitive to the accumulative error of other protocols. To this end, we create testing frame pairs of increasing difficulty and measure the pose estimation error separately for each of them. This approach gives new insights and allows to accurately study the performance of each feature or method without employing a full tracking pipeline. Following this protocol, we evaluate various directional distances in the context of silhouette-based 3d hand tracking, expressed as special cases of a generalized Chamfer distance form. An appropriate parameter setup is proposed for each of them, and a comparative study reveals the best performing method in this context.
We present a markerless motion capture approach that reconstructs the skeletal motion and detailed time-varying surface geometry of two closely interacting people from multi-view video. Due to ambiguities in feature-to-person assignments and frequent occlusions, it is not feasible to directly apply single-person capture approaches to the multi-person case. We therefore propose a combined image segmentation and tracking approach to overcome these difficulties. A new probabilistic shape and appearance model is employed to segment the input images and to assign each pixel uniquely to one person. Thereafter, a single-person markerless motion and surface capture approach can be applied to each individual, either one-by-one or in parallel, even under strong occlusions. We demonstrate the performance of our approach on several challenging multi-person motions, including dance and martial arts, and also provide a reference dataset for multi-person motion capture with ground truth.
Due to occlusions, the estimation of the full pose of a human hand interacting with an object is much more challenging than pose recovery of a hand observed in isolation. In this work we formulate an optimization problem whose solution is the 26-DOF hand pose together with the pose and model parameters of the manipulated object. Optimization seeks for the joint hand-object model that (a) best explains the incompleteness of observations resulting from occlusions due to hand-object interaction and (b) is physically plausible in the sense that the hand does not share the same physical space with the object. The proposed method is the first that solves efficiently the continuous, full-DOF, joint hand-object tracking problem based solely on markerless multicamera input. Additionally, it is the first to demonstrate how hand-object interaction can be exploited as a context that facilitates hand pose estimation, instead of being considered as a complicating factor. Extensive quantitative and qualitative experiments with simulated and real world image sequences as well as a comparative evaluation with a state-of-the-art method for pose estimation of isolated hands, support the above findings.
We present a method for tracking a hand while it is interacting with an object. This setting is arguably the one where hand-tracking has most practical relevance, but poses significant additional challenges: strong occlusions by the object as well as self-occlusions are the norm, and classical anatomical constraints need to be softened due to the external forces between hand and object. To achieve robustness to partial occlusions, we use an individual local tracker for each segment of the articulated structure. The segments are connected in a pairwise Markov random field, which enforces the anatomical hand structure through soft constraints on the joints between adjacent segments. The most likely hand configuration is found with belief propagation. Both range and color data are used as input. Experiments are presented for synthetic data with ground truth and for real data of people manipulating objects.
While research on articulated human motion and pose estimation has progressed rapidly in the last few years, there has been no systematic quantitative evaluation of competing methods to establish the current state of the art. We present data obtained using a hardware system that is able to capture synchronized video and ground-truth 3D motion. The resulting HUMANEVA datasets contain multiple subjects performing a set of predefined actions with a number of repetitions. On the order of 40,000 frames of synchronized motion capture and multi-view video (resulting in over one quarter million image frames in total ) were collected at 60 Hz with an additional 37,000 time instants of pure motion capture data. A standard set of error measures is defined for evaluating both 2D and D pose estimation and tracking algorithms. We also describe a baseline algorithm for 3D articulated tracking that uses a relatively standard Bayesian framework with optimization in the form of Sequential Importance Resampling and Annealed Particle Filtering. In the context of this baselin e algorithm we explore a variety of likelihood functions, prior models of human motion and the effects of algorithm parameters. Our experiments suggest that image observation models and motion priors play important roles in performance, and that in a multi-view laboratory environment, where initialization is available, Bayesian filtering tend s to perform well. The datasets and the software are made available to the research community. This infrastructure will support the development of new articulated motion and pose estimation algorithms, will provide a baseline for the evaluation and comparison of new methods, and will help establish the current state of the art in human pose estimati on and tracking.
Realtime multi-person 2D pose estimation is a key component in enabling machines to have an understanding of people in images and videos. In this work, we present a realtime approach to detect the 2D pose of multiple people in an image. The proposed method uses a nonparametric representation, which we refer to as Part Affinity Fields (PAFs), to learn to associate body parts with individuals in the image. This bottom-up system achieves high accuracy and realtime performance, regardless of the number of people in the image. In previous work, PAFs and body part location estimation were refined simultaneously across training stages. We demonstrate that using a PAF-only refinement is able to achieve a substantial increase in both runtime performance and accuracy. We also present the first combined body and foot keypoint detector, based on an annotated foot dataset that we have publicly released. We show that the combined detector not only reduces the inference time compared to running them sequentially, but also maintains the accuracy of each component individually. This work has culminated in the release of OpenPose, the first open-source realtime system for multi-person 2D pose detection, including body, foot, hand, and facial keypoints.
Multi-person pose estimation is fundamental to many computer vision tasks and has made significant progress in recent years. However, few previous methods explored the problem of pose estimation in crowded scenes while it remains challenging and inevitable in many scenarios. Moreover, current benchmarks cannot provide an appropriate evaluation for such cases. In this paper, we propose a novel and efficient method to tackle the problem of pose estimation in the crowd and a new dataset to better evaluate algorithms. Our model consists of two key components: joint-candidate single person pose estimation (SPPE) and global maximum joints association. With multi-peak prediction for each joint and global association using the graph model, our method is robust to inevitable interference in crowded scenes and very efficient in inference. The proposed method surpasses the state-of-the-art methods on CrowdPose dataset by 5.2 mAP and results on MSCOCO dataset demonstrate the generalization ability of our method. Source code and dataset are available at https://github.com/Jeff-sjtu/CrowdPose
We present a unified deformation model for the markerless capture of multiple scales of human movement, including facial expressions, body motion, and hand gestures. An initial model is generated by locally stitching together models of the individual parts of the human body, which we refer to as the "Frankenstein" model. This model enables the full expression of part movements, including face and hands by a single seamless model. Using a large-scale capture of people wearing everyday clothes, we optimize the Frankenstein model to create "Adam". Adam is a calibrated model that shares the same skeleton hierarchy as the initial model but can express hair and clothing geometry, making it directly usable for fitting people as they normally appear in everyday life. Finally, we demonstrate the use of these models for total motion tracking, simultaneously capturing the large-scale body movements and the subtle face and hand motion of a social group of people.
In this paper we introduce a novel dataset, the Multimodal Human-Human-Robot-Interactions (MHHRI) dataset, with the aim of studying personality simultaneously in human-human interactions (HHI) and human-robot interactions (HRI) and its relationship with engagement. Multimodal data was collected during a controlled interaction study where dyadic interactions between two human participants and triadic interactions between two human participants and a robot took place with interactants asking a set of personal questions to each other. Interactions were recorded using two static and two dynamic cameras as well as two biosensors, and meta-data was collected by having participants to fill in two types of questionnaires, for assessing their own personality traits and their perceived engagement with their partners (self labels) and for assessing personality traits of the other participants partaking in the study (acquaintance labels). As a proof of concept, we present baseline results for personality and engagement classification. Our results show that (i) trends in personality classification performance remain the same with respect to the self and the acquaintance labels across the HHI and HRI settings; (ii) for extroversion, the acquaintance labels yield better results as compared to the self labels; (iii) in general, multi-modality yields better performance for the classification of personality traits.
With the Yahoo Flickr Creative Commons 100 Million (YFCC100m) dataset, a novel dataset was introduced to the computer vision and multimedia research community. To maximize the benefit for the research community and utilize its potential, this dataset has to be made accessible by tools allowing to search for target concepts within the dataset and mechanism to browse images and videos of the dataset. Following best practice from data collections, such as ImageNet and MS COCO, this paper presents means of accessibility for the YFCC100m dataset. This includes a global analysis of the dataset and an online browser to explore and investigate subsets of the dataset in real-time. Providing statistics of the queried images and videos will enable researchers to refine their query successively, such that the users desired subset of interest can be narrowed down quickly. The final set of image and video can be downloaded as URLs from the browser for further processing.
We created the Yahoo Flickr Creative Commons 100 Million Dataseta (YFCC100M) in 2014 as part of the Yahoo Webscope program, which is a reference library of interesting and scientifically useful datasets. The YFCC100M is the largest public multimedia collection ever released, with a total of 100 million media objects, of which approximately 99.2 million are photos and 0.8 million are videos, all uploaded to Flickr between 2004 and 2014 and published under a CC commercial or noncommercial license. The dataset is distributed through Amazon Web Services as a 12.5GB compressed archive containing only metadata. However, as with many datasets, the YFCC100M is constantly evolving; over time, we have released and will continue to release various expansion packs containing data not yet in the collection; for instance, the actual photos and videos, as well as several visual and aural features extracted from the data, have already been uploaded to the cloud, ensuring the dataset remains accessible and intact for years to come. The YFCC100M dataset overcomes many of the issues affecting existing multimedia datasets in terms of modalities, metadata, licensing, and, principally, volume.
Deeper neural networks are more difficult to train. We present a residual
learning framework to ease the training of networks that are substantially
deeper than those used previously. We explicitly reformulate the layers as
learning residual functions with reference to the layer inputs, instead of
learning unreferenced functions. We provide comprehensive empirical evidence
showing that these residual networks are easier to optimize, and can gain
accuracy from considerably increased depth. On the ImageNet dataset we evaluate
residual nets with a depth of up to 152 layers---8x deeper than VGG nets but
still having lower complexity. An ensemble of these residual nets achieves
3.57% error on the ImageNet test set. This result won the 1st place on the
ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100
and 1000 layers.
The depth of representations is of central importance for many visual
recognition tasks. Solely due to our extremely deep representations, we obtain
a 28% relative improvement on the COCO object detection dataset. Deep residual
nets are foundations of our submissions to ILSVRC & COCO 2015 competitions,
where we also won the 1st places on the tasks of ImageNet detection, ImageNet
localization, COCO detection, and COCO segmentation.
Solid shapes in computer graphics are often represented with boundary descriptions, e.g. triangle meshes, but animation, physically-based simulation, and geometry processing are more realistic and accurate when explicit volume representations are available. Tetrahedral meshes which exactly contain (interpolate) the input boundary description are desirable but difficult to construct for a large class of input meshes. Character meshes and CAD models are often composed of many connected components with numerous self-intersections, non-manifold pieces, and open boundaries, precluding existing meshing algorithms. We propose an automatic algorithm handling all of these issues, resulting in a compact discretization of the input's inner volume. We only require reasonably consistent orientation of the input triangle mesh. By generalizing the winding number for arbitrary triangle meshes, we define a function that is a perfect segmentation for watertight input and is well-behaved otherwise. This function guides a graphcut segmentation of a constrained Delaunay tessellation (CDT), providing a minimal description that meets the boundary exactly and may be fed as input to existing tools to achieve element quality. We highlight our robustness on a number of examples and show applications of solving PDEs, volumetric texturing and elastic simulation.
We introduce a new dataset, Human3.6M, of 3.6 Million 3D Human poses, acquired by recording the performance of 11 subjects, under 4 different viewpoints, for training realistic human sensing systems and for evaluating the next generation of human pose estimation models. Besides increasing the size the current state of the art datasets by several orders of magnitude, we aim to complement such datasets with a diverse set of poses encountered in typical human activities (taking photos, posing, greeting, eating, etc.), with synchronized image, motion capture and depth data, and with accurate 3D body scans of all subjects involved. We also provide mixed reality videos where 3D human models are animated using motion capture data and inserted using correct 3D geometry, in complex real environments, viewed with moving cameras, and under occlusion. Finally, we provide large scale statistical models and detailed evaluation baselines for the dataset illustrating its diversity and the scope for improvement by future work in the research community. The dataset and code for the associated large-scale learning models, features, visualization tools, as well as the evaluation server, are available online at http://vision.imar.ro/human3.6m.
Capturing the skeleton motion and detailed time-varying surface geometry of multiple, closely interacting peoples is a very challenging task, even in a multicamera setup, due to frequent occlusions and ambiguities in feature-to-person assignments. To address this task, we propose a framework that exploits multiview image segmentation. To this end, a probabilistic shape and appearance model is employed to segment the input images and to assign each pixel uniquely to one person. Given the articulated template models of each person and the labeled pixels, a combined optimization scheme, which splits the skeleton pose optimization problem into a local one and a lower dimensional global one, is applied one by one to each individual, followed with surface estimation to capture detailed nonrigid deformations. We show on various sequences that our approach can capture the 3D motion of humans accurately even if they move rapidly, if they wear wide apparel, and if they are engaged in challenging multiperson motions, including dancing, wrestling, and hugging.