Lourdes Agapito

Mel Vecerik, Carl Doersch, Yi Yang et al. · deepmind

For robots to be useful outside labs and specialized factories we need a way to teach them new useful behaviors quickly. Current approaches lack either the generality to onboard new tasks without task-specific engineering, or else lack the data-efficiency to do so in an amount of time that enables practical use. In this work we explore dense tracking as a representational vehicle to allow faster and more general learning from demonstration. Our approach utilizes Track-Any-Point (TAP) models to isolate the relevant motion in a demonstration, and parameterize a low-level controller to reproduce this motion across changes in the scene configuration. We show this results in robust robot policies that can solve complex object-arrangement tasks such as shape-matching, stacking, and even full path-following tasks such as applying glue and sticking objects together, all from demonstrations that can be collected in minutes.

Hengyi Wang, Jingwen Wang, Lourdes Agapito · tencent-ai

We present Co-SLAM, a neural RGB-D SLAM system based on a hybrid representation, that performs robust camera tracking and high-fidelity surface reconstruction in real time. Co-SLAM represents the scene as a multi-resolution hash-grid to exploit its high convergence speed and ability to represent high-frequency local features. In addition, Co-SLAM incorporates one-blob encoding, to encourage surface coherence and completion in unobserved areas. This joint parametric-coordinate encoding enables real-time and robust performance by bringing the best of both worlds: fast convergence and surface hole filling. Moreover, our ray sampling strategy allows Co-SLAM to perform global bundle adjustment over all keyframes instead of requiring keyframe selection to maintain a small number of active keyframes as competing neural SLAM approaches do. Experimental results show that Co-SLAM runs at 10-17Hz and achieves state-of-the-art scene reconstruction results, and competitive tracking performance in various datasets and benchmarks (ScanNet, TUM, Replica, Synthetic RGBD). Project page: https://hengyiwang.github.io/projects/CoSLAM

Jingwen Wang, Tymoteusz Bleja, Lourdes Agapito · tencent-ai

We present GO-Surf, a direct feature grid optimization method for accurate and fast surface reconstruction from RGB-D sequences. We model the underlying scene with a learned hierarchical feature voxel grid that encapsulates multi-level geometric and appearance local information. Feature vectors are directly optimized such that after being tri-linearly interpolated, decoded by two shallow MLPs into signed distance and radiance values, and rendered via surface volume rendering, the discrepancy between synthesized and observed RGB/depth values is minimized. Our supervision signals -- RGB, depth and approximate SDF -- can be obtained directly from input images without any need for fusion or post-processing. We formulate a novel SDF gradient regularization term that encourages surface smoothness and hole filling while maintaining high frequency details. GO-Surf can optimize sequences of $1$-$2$K frames in $15$-$45$ minutes, a speedup of $\times60$ over NeuralRGB-D, the most related approach based on an MLP representation, while maintaining on par performance on standard benchmarks. Project page: https://jingwenwang95.github.io/go_surf/

Simon Giebenhain, Tobias Kirschstein, Markos Georgopoulos et al.

We propose a novel 3D morphable model for complete human heads based on hybrid neural fields. At the core of our model lies a neural parametric representation that disentangles identity and expressions in disjoint latent spaces. To this end, we capture a person's identity in a canonical space as a signed distance field (SDF), and model facial expressions with a neural deformation field. In addition, our representation achieves high-fidelity local detail by introducing an ensemble of local fields centered around facial anchor points. To facilitate generalization, we train our model on a newly-captured dataset of over 5200 head scans from 255 different identities using a custom high-end 3D scanning setup. Our dataset significantly exceeds comparable existing datasets, both with respect to quality and completeness of geometry, averaging around 3.5M mesh faces per scan. Finally, we demonstrate that our approach outperforms state-of-the-art methods in terms of fitting error and reconstruction quality.

Jingwen Wang, Juan Tarrio, Lourdes Agapito et al. · tencent-ai

The availability of real-time semantics greatly improves the core geometric functionality of SLAM systems, enabling numerous robotic and AR/VR applications. We present a new methodology for real-time semantic mapping from RGB-D sequences that combines a 2D neural network and a 3D network based on a SLAM system with 3D occupancy mapping. When segmenting a new frame we perform latent feature re-projection from previous frames based on differentiable rendering. Fusing re-projected feature maps from previous frames with current-frame features greatly improves image segmentation quality, compared to a baseline that processes images independently. For 3D map processing, we propose a novel geometric quasi-planar over-segmentation method that groups 3D map elements likely to belong to the same semantic classes, relying on surface normals. We also describe a novel neural network design for lightweight semantic map post-processing. Our system achieves state-of-the-art semantic mapping quality within 2D-3D networks-based systems and matches the performance of 3D convolutional networks on three real indoor datasets, while working in real-time. Moreover, it shows better cross-sensor generalization abilities compared to 3D CNNs, enabling training and inference with different depth sensors. Code and data will be released on project page: http://jingwenwang95.github.io/SeMLaPS

Anita Rau, Binod Bhattarai, Lourdes Agapito et al.

Deducing the 3D structure of endoscopic scenes from images is exceedingly challenging. In addition to deformation and view-dependent lighting, tubular structures like the colon present problems stemming from their self-occluding and repetitive anatomical structure. In this paper, we propose SimCol, a synthetic dataset for camera pose estimation in colonoscopy, and a novel method that explicitly learns a bimodal distribution to predict the endoscope pose. Our dataset replicates real colonoscope motion and highlights the drawbacks of existing methods. We publish 18k RGB images from simulated colonoscopy with corresponding depth and camera poses and make our data generation environment in Unity publicly available. We evaluate different camera pose prediction methods and demonstrate that, when trained on our data, they generalize to real colonoscopy sequences, and our bimodal approach outperforms prior unimodal work.

Denis Hadjivelichkov, Sicelukwanda Zwane, Marc Peter Deisenroth et al.

In this work, we tackle one-shot visual search of object parts. Given a single reference image of an object with annotated affordance regions, we segment semantically corresponding parts within a target scene. We propose AffCorrs, an unsupervised model that combines the properties of pre-trained DINO-ViT's image descriptors and cyclic correspondences. We use AffCorrs to find corresponding affordances both for intra- and inter-class one-shot part segmentation. This task is more difficult than supervised alternatives, but enables future work such as learning affordances via imitation and assisted teleoperation.

Anita Rau, Binod Bhattarai, Lourdes Agapito et al.

Colorectal cancer remains one of the deadliest cancers in the world. In recent years computer-aided methods have aimed to enhance cancer screening and improve the quality and availability of colonoscopies by automatizing sub-tasks. One such task is predicting depth from monocular video frames, which can assist endoscopic navigation. As ground truth depth from standard in-vivo colonoscopy remains unobtainable due to hardware constraints, two approaches have aimed to circumvent the need for real training data: supervised methods trained on labeled synthetic data and self-supervised models trained on unlabeled real data. However, self-supervised methods depend on unreliable loss functions that struggle with edges, self-occlusion, and lighting inconsistency. Methods trained on synthetic data can provide accurate depth for synthetic geometries but do not use any geometric supervisory signal from real data and overfit to synthetic anatomies and properties. This work proposes a novel approach to leverage labeled synthetic and unlabeled real data. While previous domain adaptation methods indiscriminately enforce the distributions of both input data modalities to coincide, we focus on the end task, depth prediction, and translate only essential information between the input domains. Our approach results in more resilient and accurate depth maps of real colonoscopy sequences.

Mirgahney Mohamed, Lourdes Agapito

We propose Geometric Neural Parametric Models (GNPM), a learned parametric model that takes into account the local structure of data to learn disentangled shape and pose latent spaces of 4D dynamics, using a geometric-aware architecture on point clouds. Temporally consistent 3D deformations are estimated without the need for dense correspondences at training time, by exploiting cycle consistency. Besides its ability to learn dense correspondences, GNPMs also enable latent-space manipulations such as interpolation and shape/pose transfer. We evaluate GNPMs on various datasets of clothed humans, and show that it achieves comparable performance to state-of-the-art methods that require dense correspondences during training.

Hengyi Wang, Lourdes Agapito

We present Spann3R, a novel approach for dense 3D reconstruction from ordered or unordered image collections. Built on the DUSt3R paradigm, Spann3R uses a transformer-based architecture to directly regress pointmaps from images without any prior knowledge of the scene or camera parameters. Unlike DUSt3R, which predicts per image-pair pointmaps each expressed in its local coordinate frame, Spann3R can predict per-image pointmaps expressed in a global coordinate system, thus eliminating the need for optimization-based global alignment. The key idea of Spann3R is to manage an external spatial memory that learns to keep track of all previous relevant 3D information. Spann3R then queries this spatial memory to predict the 3D structure of the next frame in a global coordinate system. Taking advantage of DUSt3R's pre-trained weights, and further fine-tuning on a subset of datasets, Spann3R shows competitive performance and generalization ability on various unseen datasets and can process ordered image collections in real time. Project page: \url{https://hengyiwang.github.io/projects/spanner}

David Svitov, Pietro Morerio, Lourdes Agapito et al.

We present a CloseUpAvatar - a novel approach for articulated human avatar representation dealing with more general camera motions, while preserving rendering quality for close-up views. CloseUpAvatar represents an avatar as a set of textured planes with two sets of learnable textures for low and high-frequency detail. The method automatically switches to high-frequency textures only for cameras positioned close to the avatar's surface and gradually reduces their impact as the camera moves farther away. Such parametrization of the avatar enables CloseUpAvatar to adjust rendering quality based on camera distance ensuring realistic rendering across a wider range of camera orientations than previous approaches. We provide experiments using the ActorsHQ dataset with high-resolution input images. CloseUpAvatar demonstrates both qualitative and quantitative improvements over existing methods in rendering from novel wide range camera positions, while maintaining high FPS by limiting the number of required primitives.

Mirgahney Mohamed, Lourdes Agapito

We propose DynamicSurf, a model-free neural implicit surface reconstruction method for high-fidelity 3D modelling of non-rigid surfaces from monocular RGB-D video. To cope with the lack of multi-view cues in monocular sequences of deforming surfaces, one of the most challenging settings for 3D reconstruction, DynamicSurf exploits depth, surface normals, and RGB losses to improve reconstruction fidelity and optimisation time. DynamicSurf learns a neural deformation field that maps a canonical representation of the surface geometry to the current frame. We depart from current neural non-rigid surface reconstruction models by designing the canonical representation as a learned feature grid which leads to faster and more accurate surface reconstruction than competing approaches that use a single MLP. We demonstrate DynamicSurf on public datasets and show that it can optimize sequences of varying frames with $6\times$ speedup over pure MLP-based approaches while achieving comparable results to the state-of-the-art methods. Project is available at https://mirgahney.github.io//DynamicSurf.io/.

Wonbong Jang, Jonathan Tremblay, Lourdes Agapito

Generating novel views of a natural scene, e.g., every-day scenes both indoors and outdoors, from a single view is an under-explored problem, even though it is an organic extension to the object-centric novel view synthesis. Existing diffusion-based approaches focus rather on small camera movements in real scenes or only consider unnatural object-centric scenes, limiting their potential applications in real-world settings. In this paper we move away from these constrained regimes and propose a 3D diffusion model trained with image-only losses on a large-scale dataset of real-world, multi-category, unaligned, and casually acquired videos of everyday scenes. We propose DT-NVS, a 3D-aware diffusion model for generalized novel view synthesis that exploits a transformer-based architecture backbone. We make significant contributions to transformer and self-attention architectures to translate images to 3d representations, and novel camera conditioning strategies to allow training on real-world unaligned datasets. In addition, we introduce a novel training paradigm swapping the role of reference frame between the conditioning image and the sampled noisy input. We evaluate our approach on the 3D task of generalized novel view synthesis from a single input image and show improvements over state-of-the-art 3D aware diffusion models and deterministic approaches, while generating diverse outputs.

Wonbong Jang, Lourdes Agapito

CodeNeRF is an implicit 3D neural representation that learns the variation of object shapes and textures across a category and can be trained, from a set of posed images, to synthesize novel views of unseen objects. Unlike the original NeRF, which is scene specific, CodeNeRF learns to disentangle shape and texture by learning separate embeddings. At test time, given a single unposed image of an unseen object, CodeNeRF jointly estimates camera viewpoint, and shape and appearance codes via optimization. Unseen objects can be reconstructed from a single image, and then rendered from new viewpoints or their shape and texture edited by varying the latent codes. We conduct experiments on the SRN benchmark, which show that CodeNeRF generalises well to unseen objects and achieves on-par performance with methods that require known camera pose at test time. Our results on real-world images demonstrate that CodeNeRF can bridge the sim-to-real gap. Project page: \url{https://github.com/wayne1123/code-nerf}

David Svitov, Pietro Morerio, Lourdes Agapito et al.

We present HAHA - a novel approach for animatable human avatar generation from monocular input videos. The proposed method relies on learning the trade-off between the use of Gaussian splatting and a textured mesh for efficient and high fidelity rendering. We demonstrate its efficiency to animate and render full-body human avatars controlled via the SMPL-X parametric model. Our model learns to apply Gaussian splatting only in areas of the SMPL-X mesh where it is necessary, like hair and out-of-mesh clothing. This results in a minimal number of Gaussians being used to represent the full avatar, and reduced rendering artifacts. This allows us to handle the animation of small body parts such as fingers that are traditionally disregarded. We demonstrate the effectiveness of our approach on two open datasets: SnapshotPeople and X-Humans. Our method demonstrates on par reconstruction quality to the state-of-the-art on SnapshotPeople, while using less than a third of Gaussians. HAHA outperforms previous state-of-the-art on novel poses from X-Humans both quantitatively and qualitatively.

Simon Giebenhain, Tobias Kirschstein, Markos Georgopoulos et al.

We present Monocular Neural Parametric Head Models (MonoNPHM) for dynamic 3D head reconstructions from monocular RGB videos. To this end, we propose a latent appearance space that parameterizes a texture field on top of a neural parametric model. We constrain predicted color values to be correlated with the underlying geometry such that gradients from RGB effectively influence latent geometry codes during inverse rendering. To increase the representational capacity of our expression space, we augment our backward deformation field with hyper-dimensions, thus improving color and geometry representation in topologically challenging expressions. Using MonoNPHM as a learned prior, we approach the task of 3D head reconstruction using signed distance field based volumetric rendering. By numerically inverting our backward deformation field, we incorporated a landmark loss using facial anchor points that are closely tied to our canonical geometry representation. To evaluate the task of dynamic face reconstruction from monocular RGB videos we record 20 challenging Kinect sequences under casual conditions. MonoNPHM outperforms all baselines with a significant margin, and makes an important step towards easily accessible neural parametric face models through RGB tracking.

Wonbong Jang, Philippe Weinzaepfel, Vincent Leroy et al.

We present Pow3r, a novel large 3D vision regression model that is highly versatile in the input modalities it accepts. Unlike previous feed-forward models that lack any mechanism to exploit known camera or scene priors at test time, Pow3r incorporates any combination of auxiliary information such as intrinsics, relative pose, dense or sparse depth, alongside input images, within a single network. Building upon the recent DUSt3R paradigm, a transformer-based architecture that leverages powerful pre-training, our lightweight and versatile conditioning acts as additional guidance for the network to predict more accurate estimates when auxiliary information is available. During training we feed the model with random subsets of modalities at each iteration, which enables the model to operate under different levels of known priors at test time. This in turn opens up new capabilities, such as performing inference in native image resolution, or point-cloud completion. Our experiments on 3D reconstruction, depth completion, multi-view depth prediction, multi-view stereo, and multi-view pose estimation tasks yield state-of-the-art results and confirm the effectiveness of Pow3r at exploiting all available information. The project webpage is https://europe.naverlabs.com/pow3r.

Wonbong Jang, Lourdes Agapito

We propose NViST, a transformer-based model for efficient and generalizable novel-view synthesis from a single image for real-world scenes. In contrast to many methods that are trained on synthetic data, object-centred scenarios, or in a category-specific manner, NViST is trained on MVImgNet, a large-scale dataset of casually-captured real-world videos of hundreds of object categories with diverse backgrounds. NViST transforms image inputs directly into a radiance field, conditioned on camera parameters via adaptive layer normalisation. In practice, NViST exploits fine-tuned masked autoencoder (MAE) features and translates them to 3D output tokens via cross-attention, while addressing occlusions with self-attention. To move away from object-centred datasets and enable full scene synthesis, NViST adopts a 6-DOF camera pose model and only requires relative pose, dropping the need for canonicalization of the training data, which removes a substantial barrier to it being used on casually captured datasets. We show results on unseen objects and categories from MVImgNet and even generalization to casual phone captures. We conduct qualitative and quantitative evaluations on MVImgNet and ShapeNet to show that our model represents a step forward towards enabling true in-the-wild generalizable novel-view synthesis from a single image. Project webpage: https://wbjang.github.io/nvist_webpage.

David Svitov, Pietro Morerio, Lourdes Agapito et al.

We present billboard Splatting (BBSplat) - a novel approach for novel view synthesis based on textured geometric primitives. BBSplat represents the scene as a set of optimizable textured planar primitives with learnable RGB textures and alpha-maps to control their shape. BBSplat primitives can be used in any Gaussian Splatting pipeline as drop-in replacements for Gaussians. The proposed primitives close the rendering quality gap between 2D and 3D Gaussian Splatting (GS), enabling the accurate extraction of 3D mesh as in the 2DGS framework. Additionally, the explicit nature of planar primitives enables the use of the ray-tracing effects in rasterization. Our novel regularization term encourages textures to have a sparser structure, enabling an efficient compression that leads to a reduction in the storage space of the model up to x17 times compared to 3DGS. Our experiments show the efficiency of BBSplat on standard datasets of real indoor and outdoor scenes such as Tanks&Temples, DTU, and Mip-NeRF-360. Namely, we achieve a state-of-the-art PSNR of 29.72 for DTU at Full HD resolution.

Simon Giebenhain, Tobias Kirschstein, Martin Rünz et al.

We address the 3D reconstruction of human faces from a single RGB image. To this end, we propose Pixel3DMM, a set of highly-generalized vision transformers which predict per-pixel geometric cues in order to constrain the optimization of a 3D morphable face model (3DMM). We exploit the latent features of the DINO foundation model, and introduce a tailored surface normal and uv-coordinate prediction head. We train our model by registering three high-quality 3D face datasets against the FLAME mesh topology, which results in a total of over 1,000 identities and 976K images. For 3D face reconstruction, we propose a FLAME fitting opitmization that solves for the 3DMM parameters from the uv-coordinate and normal estimates. To evaluate our method, we introduce a new benchmark for single-image face reconstruction, which features high diversity facial expressions, viewing angles, and ethnicities. Crucially, our benchmark is the first to evaluate both posed and neutral facial geometry. Ultimately, our method outperforms the most competitive baselines by over 15% in terms of geometric accuracy for posed facial expressions.

Shalini Maiti, Lourdes Agapito, Filippos Kokkinos

Rapid advancements in text-to-3D generation require robust and scalable evaluation metrics that align closely with human judgment, a need unmet by current metrics such as PSNR and CLIP, which require ground-truth data or focus only on prompt fidelity. To address this, we introduce Gen3DEval, a novel evaluation framework that leverages vision large language models (vLLMs) specifically fine-tuned for 3D object quality assessment. Gen3DEval evaluates text fidelity, appearance, and surface quality by analyzing 3D surface normals, without requiring ground-truth comparisons, bridging the gap between automated metrics and user preferences. Compared to state-of-the-art task-agnostic models, Gen3DEval demonstrates superior performance in user-aligned evaluations, placing it as a comprehensive and accessible benchmark for future research on text-to-3D generation. The project page can be found here: \href{https://shalini-maiti.github.io/gen3deval.github.io/}{https://shalini-maiti.github.io/gen3deval.github.io/}.

Hengyi Wang, Lourdes Agapito

We present AMB3R, a multi-view feed-forward model for dense 3D reconstruction on a metric-scale that addresses diverse 3D vision tasks. The key idea is to leverage a sparse, yet compact, volumetric scene representation as our backend, enabling geometric reasoning with spatial compactness. Although trained solely for multi-view reconstruction, we demonstrate that AMB3R can be seamlessly extended to uncalibrated visual odometry (online) or large-scale structure from motion without the need for task-specific fine-tuning or test-time optimization. Compared to prior pointmap-based models, our approach achieves state-of-the-art performance in camera pose, depth, and metric-scale estimation, 3D reconstruction, and even surpasses optimization-based SLAM and SfM methods with dense reconstruction priors on common benchmarks.

Shalini Maiti, Lourdes Agapito, Benjamin Graham

For non-rigid objects, predicting the 3D shape from 2D keypoint observations is ill-posed due to occlusions, and the need to disentangle changes in viewpoint and changes in shape. This challenge has often been addressed by embedding low-rank constraints into specialized models. These models can be hard to train, as they depend on finding a canonical way of aligning observations, before they can learn detailed geometry. These constraints have limited the reconstruction quality. We show that generic, high capacity models, trained with an unsupervised loss, allow for more accurate predicted shapes. In particular, applying low-rank constraints to localized subsets of the full shape allows the high capacity to be suitably constrained. We reduce the state-of-the-art reconstruction error on the S-Up3D dataset by over 70%.

Mirgahney Mohamed, Harry Jake Cunningham, Marc P. Deisenroth et al.

Human motion generation has paramount importance in computer animation. It is a challenging generative temporal modelling task due to the vast possibilities of human motion, high human sensitivity to motion coherence and the difficulty of accurately generating fine-grained motions. Recently, diffusion methods have been proposed for human motion generation due to their high sample quality and expressiveness. However, generated sequences still suffer from motion incoherence, and are limited to short duration, and simpler motion and take considerable time during inference. To address these limitations, we propose \textit{RecMoDiffuse: Recurrent Flow Diffusion}, a new recurrent diffusion formulation for temporal modelling. Unlike previous work, which applies diffusion to the whole sequence without any temporal dependency, an approach that inherently makes temporal consistency hard to achieve. Our method explicitly enforces temporal constraints with the means of normalizing flow models in the diffusion process and thereby extends diffusion to the temporal dimension. We demonstrate the effectiveness of RecMoDiffuse in the temporal modelling of human motion. Our experiments show that RecMoDiffuse achieves comparable results with state-of-the-art methods while generating coherent motion sequences and reducing the computational overhead in the inference stage.

Mustafa Işık, Martin Rünz, Markos Georgopoulos et al.

Representing human performance at high-fidelity is an essential building block in diverse applications, such as film production, computer games or videoconferencing. To close the gap to production-level quality, we introduce HumanRF, a 4D dynamic neural scene representation that captures full-body appearance in motion from multi-view video input, and enables playback from novel, unseen viewpoints. Our novel representation acts as a dynamic video encoding that captures fine details at high compression rates by factorizing space-time into a temporal matrix-vector decomposition. This allows us to obtain temporally coherent reconstructions of human actors for long sequences, while representing high-resolution details even in the context of challenging motion. While most research focuses on synthesizing at resolutions of 4MP or lower, we address the challenge of operating at 12MP. To this end, we introduce ActorsHQ, a novel multi-view dataset that provides 12MP footage from 160 cameras for 16 sequences with high-fidelity, per-frame mesh reconstructions. We demonstrate challenges that emerge from using such high-resolution data and show that our newly introduced HumanRF effectively leverages this data, making a significant step towards production-level quality novel view synthesis.

Mel Vecerik, Jackie Kay, Raia Hadsell et al.

Dense object tracking, the ability to localize specific object points with pixel-level accuracy, is an important computer vision task with numerous downstream applications in robotics. Existing approaches either compute dense keypoint embeddings in a single forward pass, meaning the model is trained to track everything at once, or allocate their full capacity to a sparse predefined set of points, trading generality for accuracy. In this paper we explore a middle ground based on the observation that the number of relevant points at a given time are typically relatively few, e.g. grasp points on a target object. Our main contribution is a novel architecture, inspired by few-shot task adaptation, which allows a sparse-style network to condition on a keypoint embedding that indicates which point to track. Our central finding is that this approach provides the generality of dense-embedding models, while offering accuracy significantly closer to sparse-keypoint approaches. We present results illustrating this capacity vs. accuracy trade-off, and demonstrate the ability to zero-shot transfer to new object instances (within-class) using a real-robot pick-and-place task.

Jingwen Wang, Martin Rünz, Lourdes Agapito

We propose DSP-SLAM, an object-oriented SLAM system that builds a rich and accurate joint map of dense 3D models for foreground objects, and sparse landmark points to represent the background. DSP-SLAM takes as input the 3D point cloud reconstructed by a feature-based SLAM system and equips it with the ability to enhance its sparse map with dense reconstructions of detected objects. Objects are detected via semantic instance segmentation, and their shape and pose is estimated using category-specific deep shape embeddings as priors, via a novel second order optimization. Our object-aware bundle adjustment builds a pose-graph to jointly optimize camera poses, object locations and feature points. DSP-SLAM can operate at 10 frames per second on 3 different input modalities: monocular, stereo, or stereo+LiDAR. We demonstrate DSP-SLAM operating at almost frame rate on monocular-RGB sequences from the Friburg and Redwood-OS datasets, and on stereo+LiDAR sequences on the KITTI odometry dataset showing that it achieves high-quality full object reconstructions, even from partial observations, while maintaining a consistent global map. Our evaluation shows improvements in object pose and shape reconstruction with respect to recent deep prior-based reconstruction methods and reductions in camera tracking drift on the KITTI dataset.

Armin Mustafa, Akin Caliskan, Lourdes Agapito et al.

We present a new end-to-end learning framework to obtain detailed and spatially coherent reconstructions of multiple people from a single image. Existing multi-person methods suffer from two main drawbacks: they are often model-based and therefore cannot capture accurate 3D models of people with loose clothing and hair; or they require manual intervention to resolve occlusions or interactions. Our method addresses both limitations by introducing the first end-to-end learning approach to perform model-free implicit reconstruction for realistic 3D capture of multiple clothed people in arbitrary poses (with occlusions) from a single image. Our network simultaneously estimates the 3D geometry of each person and their 6DOF spatial locations, to obtain a coherent multi-human reconstruction. In addition, we introduce a new synthetic dataset that depicts images with a varying number of inter-occluded humans and a variety of clothing and hair styles. We demonstrate robust, high-resolution reconstructions on images of multiple humans with complex occlusions, loose clothing and a large variety of poses and scenes. Our quantitative evaluation on both synthetic and real-world datasets demonstrates state-of-the-art performance with significant improvements in the accuracy and completeness of the reconstructions over competing approaches.

Denis Tome, Thiemo Alldieck, Patrick Peluse et al.

We present a solution to egocentric 3D body pose estimation from monocular images captured from downward looking fish-eye cameras installed on the rim of a head mounted VR device. This unusual viewpoint leads to images with unique visual appearance, with severe self-occlusions and perspective distortions that result in drastic differences in resolution between lower and upper body. We propose an encoder-decoder architecture with a novel multi-branch decoder designed to account for the varying uncertainty in 2D predictions. The quantitative evaluation, on synthetic and real-world datasets, shows that our strategy leads to substantial improvements in accuracy over state of the art egocentric approaches. To tackle the lack of labelled data we also introduced a large photo-realistic synthetic dataset. xR-EgoPose offers high quality renderings of people with diverse skintones, body shapes and clothing, performing a range of actions. Our experiments show that the high variability in our new synthetic training corpus leads to good generalization to real world footage and to state of theart results on real world datasets with ground truth. Moreover, an evaluation on the Human3.6M benchmark shows that the performance of our method is on par with top performing approaches on the more classic problem of 3D human pose from a third person viewpoint.

Mel Vecerik, Jean-Baptiste Regli, Oleg Sushkov et al.

A robot's ability to act is fundamentally constrained by what it can perceive. Many existing approaches to visual representation learning utilize general-purpose training criteria, e.g. image reconstruction, smoothness in latent space, or usefulness for control, or else make use of large datasets annotated with specific features (bounding boxes, segmentations, etc.). However, both approaches often struggle to capture the fine-detail required for precision tasks on specific objects, e.g. grasping and mating a plug and socket. We argue that these difficulties arise from a lack of geometric structure in these models. In this work we advocate semantic 3D keypoints as a visual representation, and present a semi-supervised training objective that can allow instance or category-level keypoints to be trained to 1-5 millimeter-accuracy with minimal supervision. Furthermore, unlike local texture-based approaches, our model integrates contextual information from a large area and is therefore robust to occlusion, noise, and lack of discernible texture. We demonstrate that this ability to locate semantic keypoints enables high level scripting of human understandable behaviours. Finally we show that these keypoints provide a good way to define reward functions for reinforcement learning and are a good representation for training agents.

Michael Firman, Neill D. F. Campbell, Lourdes Agapito et al.

Many structured prediction tasks in machine vision have a collection of acceptable answers, instead of one definitive ground truth answer. Segmentation of images, for example, is subject to human labeling bias. Similarly, there are multiple possible pixel values that could plausibly complete occluded image regions. State-of-the art supervised learning methods are typically optimized to make a single test-time prediction for each query, failing to find other modes in the output space. Existing methods that allow for sampling often sacrifice speed or accuracy. We introduce a simple method for training a neural network, which enables diverse structured predictions to be made for each test-time query. For a single input, we learn to predict a range of possible answers. We compare favorably to methods that seek diversity through an ensemble of networks. Such stochastic multiple choice learning faces mode collapse, where one or more ensemble members fail to receive any training signal. Our best performing solution can be deployed for various tasks, and just involves small modifications to the existing single-mode architecture, loss function, and training regime. We demonstrate that our method results in quantitative improvements across three challenging tasks: 2D image completion, 3D volume estimation, and flow prediction.

Kejie Li, Martin Rünz, Meng Tang et al.

Object-oriented maps are important for scene understanding since they jointly capture geometry and semantics, allow individual instantiation and meaningful reasoning about objects. We introduce FroDO, a method for accurate 3D reconstruction of object instances from RGB video that infers object location, pose and shape in a coarse-to-fine manner. Key to FroDO is to embed object shapes in a novel learnt space that allows seamless switching between sparse point cloud and dense DeepSDF decoding. Given an input sequence of localized RGB frames, FroDO first aggregates 2D detections to instantiate a category-aware 3D bounding box per object. A shape code is regressed using an encoder network before optimizing shape and pose further under the learnt shape priors using sparse and dense shape representations. The optimization uses multi-view geometric, photometric and silhouette losses. We evaluate on real-world datasets, including Pix3D, Redwood-OS, and ScanNet, for single-view, multi-view, and multi-object reconstruction.

Denis Tome, Patrick Peluse, Lourdes Agapito et al.

We present a new solution to egocentric 3D body pose estimation from monocular images captured from a downward looking fish-eye camera installed on the rim of a head mounted virtual reality device. This unusual viewpoint, just 2 cm. away from the user's face, leads to images with unique visual appearance, characterized by severe self-occlusions and strong perspective distortions that result in a drastic difference in resolution between lower and upper body. Our contribution is two-fold. Firstly, we propose a new encoder-decoder architecture with a novel dual branch decoder designed specifically to account for the varying uncertainty in the 2D joint locations. Our quantitative evaluation, both on synthetic and real-world datasets, shows that our strategy leads to substantial improvements in accuracy over state of the art egocentric pose estimation approaches. Our second contribution is a new large-scale photorealistic synthetic dataset - xR-EgoPose - offering 383K frames of high quality renderings of people with a diversity of skin tones, body shapes, clothing, in a variety of backgrounds and lighting conditions, performing a range of actions. Our experiments show that the high variability in our new synthetic training corpus leads to good generalization to real world footage and to state of the art results on real world datasets with ground truth. Moreover, an evaluation on the Human3.6M benchmark shows that the performance of our method is on par with top performing approaches on the more classic problem of 3D human pose from a third person viewpoint.

Adrian Penate-Sanchez, Lourdes Agapito

We present 3D Pick & Mix, a new 3D shape retrieval system that provides users with a new level of freedom to explore 3D shape and Internet image collections by introducing the ability to reason about objects at the level of their constituent parts. While classic retrieval systems can only formulate simple searches such as "find the 3D model that is most similar to the input image" our new approach can formulate advanced and semantically meaningful search queries such as: "find me the 3D model that best combines the design of the legs of the chair in image 1 but with no armrests, like the chair in image 2". Many applications could benefit from such rich queries, users could browse through catalogues of furniture and pick and mix parts, combining for example the legs of a chair from one shop and the armrests from another shop.

Denis Tome, Matteo Toso, Lourdes Agapito et al.

We propose a CNN-based approach for multi-camera markerless motion capture of the human body. Unlike existing methods that first perform pose estimation on individual cameras and generate 3D models as post-processing, our approach makes use of 3D reasoning throughout a multi-stage approach. This novelty allows us to use provisional 3D models of human pose to rethink where the joints should be located in the image and to recover from past mistakes. Our principled refinement of 3D human poses lets us make use of image cues, even from images where we previously misdetected joints, to refine our estimates as part of an end-to-end approach. Finally, we demonstrate how the high-quality output of our multi-camera setup can be used as an additional training source to improve the accuracy of existing single camera models.

Martin Rünz, Maud Buffier, Lourdes Agapito

We present MaskFusion, a real-time, object-aware, semantic and dynamic RGB-D SLAM system that goes beyond traditional systems which output a purely geometric map of a static scene. MaskFusion recognizes, segments and assigns semantic class labels to different objects in the scene, while tracking and reconstructing them even when they move independently from the camera. As an RGB-D camera scans a cluttered scene, image-based instance-level semantic segmentation creates semantic object masks that enable real-time object recognition and the creation of an object-level representation for the world map. Unlike previous recognition-based SLAM systems, MaskFusion does not require known models of the objects it can recognize, and can deal with multiple independent motions. MaskFusion takes full advantage of using instance-level semantic segmentation to enable semantic labels to be fused into an object-aware map, unlike recent semantics enabled SLAM systems that perform voxel-level semantic segmentation. We show augmented-reality applications that demonstrate the unique features of the map output by MaskFusion: instance-aware, semantic and dynamic.

Garoe Dorta, Sara Vicente, Lourdes Agapito et al.

Variational auto-encoders (VAEs) are a popular and powerful deep generative model. Previous works on VAEs have assumed a factorized likelihood model, whereby the output uncertainty of each pixel is assumed to be independent. This approximation is clearly limited as demonstrated by observing a residual image from a VAE reconstruction, which often possess a high level of structure. This paper demonstrates a novel scheme to incorporate a structured Gaussian likelihood prediction network within the VAE that allows the residual correlations to be modeled. Our novel architecture, with minimal increase in complexity, incorporates the covariance matrix prediction within the VAE. We also propose a new mechanism for allowing structured uncertainty on color images. Furthermore, we provide a scheme for effectively training this model, and include some suggestions for improving performance in terms of efficiency or modeling longer range correlations.

Garoe Dorta, Sara Vicente, Lourdes Agapito et al.

This paper is the first work to propose a network to predict a structured uncertainty distribution for a synthesized image. Previous approaches have been mostly limited to predicting diagonal covariance matrices. Our novel model learns to predict a full Gaussian covariance matrix for each reconstruction, which permits efficient sampling and likelihood evaluation. We demonstrate that our model can accurately reconstruct ground truth correlated residual distributions for synthetic datasets and generate plausible high frequency samples for real face images. We also illustrate the use of these predicted covariances for structure preserving image denoising.

Qi Liu-Yin, Rui Yu, Lourdes Agapito et al.

We demonstrate the use of shape-from-shading (SfS) to improve both the quality and the robustness of 3D reconstruction of dynamic objects captured by a single camera. Unlike previous approaches that made use of SfS as a post-processing step, we offer a principled integrated approach that solves dynamic object tracking and reconstruction and SfS as a single unified cost function. Moving beyond Lambertian S f S , we propose a general approach that models both specularities and shading while simultaneously tracking and reconstructing general dynamic objects. Solving these problems jointly prevents the kinds of tracking failures which can not be recovered from by pipeline approaches. We show state-of-the-art results both qualitatively and quantitatively.

Martin Rünz, Lourdes Agapito

In this paper we introduce Co-Fusion, a dense SLAM system that takes a live stream of RGB-D images as input and segments the scene into different objects (using either motion or semantic cues) while simultaneously tracking and reconstructing their 3D shape in real time. We use a multiple model fitting approach where each object can move independently from the background and still be effectively tracked and its shape fused over time using only the information from pixels associated with that object label. Previous attempts to deal with dynamic scenes have typically considered moving regions as outliers, and consequently do not model their shape or track their motion over time. In contrast, we enable the robot to maintain 3D models for each of the segmented objects and to improve them over time through fusion. As a result, our system can enable a robot to maintain a scene description at the object level which has the potential to allow interactions with its working environment; even in the case of dynamic scenes.

Denis Tome, Chris Russell, Lourdes Agapito

We propose a unified formulation for the problem of 3D human pose estimation from a single raw RGB image that reasons jointly about 2D joint estimation and 3D pose reconstruction to improve both tasks. We take an integrated approach that fuses probabilistic knowledge of 3D human pose with a multi-stage CNN architecture and uses the knowledge of plausible 3D landmark locations to refine the search for better 2D locations. The entire process is trained end-to-end, is extremely efficient and obtains state- of-the-art results on Human3.6M outperforming previous approaches both on 2D and 3D errors.

Rui Yu, Chris Russell, Lourdes Agapito

We propose a novel Linear Program (LP) based formula- tion for solving jigsaw puzzles. We formulate jigsaw solving as a set of successive global convex relaxations of the stan- dard NP-hard formulation, that can describe both jigsaws with pieces of unknown position and puzzles of unknown po- sition and orientation. The main contribution and strength of our approach comes from the LP assembly strategy. In contrast to existing greedy methods, our LP solver exploits all the pairwise matches simultaneously, and computes the position of each piece/component globally. The main ad- vantages of our LP approach include: (i) a reduced sensi- tivity to local minima compared to greedy approaches, since our successive approximations are global and convex and (ii) an increased robustness to the presence of mismatches in the pairwise matches due to the use of a weighted L1 penalty. To demonstrate the effectiveness of our approach, we test our algorithm on public jigsaw datasets and show that it outperforms state-of-the-art methods.

Joao Carreira, Sara Vicente, Lourdes Agapito et al.

While data has certainly taken the center stage in computer vision in recent years, it can still be difficult to obtain in certain scenarios. In particular, acquiring ground truth 3D shapes of objects pictured in 2D images remains a challenging feat and this has hampered progress in recognition-based object reconstruction from a single image. Here we propose to bypass previous solutions such as 3D scanning or manual design, that scale poorly, and instead populate object category detection datasets semi-automatically with dense, per-object 3D reconstructions, bootstrapped from:(i) class labels, (ii) ground truth figure-ground segmentations and (iii) a small set of keypoint annotations. Our proposed algorithm first estimates camera viewpoint using rigid structure-from-motion and then reconstructs object shapes by optimizing over visual hull proposals guided by loose within-class shape similarity assumptions. The visual hull sampling process attempts to intersect an object's projection cone with the cones of minimal subsets of other similar objects among those pictured from certain vantage points. We show that our method is able to produce convincing per-object 3D reconstructions and to accurately estimate cameras viewpoints on one of the most challenging existing object-category detection datasets, PASCAL VOC. We hope that our results will re-stimulate interest on joint object recognition and 3D reconstruction from a single image.