Ulrich Neumann

Quankai Gao, Qiangeng Xu, Hao Su et al.

We propose Strivec, a novel neural representation that models a 3D scene as a radiance field with sparsely distributed and compactly factorized local tensor feature grids. Our approach leverages tensor decomposition, following the recent work TensoRF, to model the tensor grids. In contrast to TensoRF which uses a global tensor and focuses on their vector-matrix decomposition, we propose to utilize a cloud of local tensors and apply the classic CANDECOMP/PARAFAC (CP) decomposition to factorize each tensor into triple vectors that express local feature distributions along spatial axes and compactly encode a local neural field. We also apply multi-scale tensor grids to discover the geometry and appearance commonalities and exploit spatial coherence with the tri-vector factorization at multiple local scales. The final radiance field properties are regressed by aggregating neural features from multiple local tensors across all scales. Our tri-vector tensors are sparsely distributed around the actual scene surface, discovered by a fast coarse reconstruction, leveraging the sparsity of a 3D scene. We demonstrate that our model can achieve better rendering quality while using significantly fewer parameters than previous methods, including TensoRF and Instant-NGP.

Yiqi Zhong, Luming Liang, Ilya Zharkov et al.

A central challenge of video prediction lies where the system has to reason the objects' future motions from image frames while simultaneously maintaining the consistency of their appearances across frames. This work introduces an end-to-end trainable two-stream video prediction framework, Motion-Matrix-based Video Prediction (MMVP), to tackle this challenge. Unlike previous methods that usually handle motion prediction and appearance maintenance within the same set of modules, MMVP decouples motion and appearance information by constructing appearance-agnostic motion matrices. The motion matrices represent the temporal similarity of each and every pair of feature patches in the input frames, and are the sole input of the motion prediction module in MMVP. This design improves video prediction in both accuracy and efficiency, and reduces the model size. Results of extensive experiments demonstrate that MMVP outperforms state-of-the-art systems on public data sets by non-negligible large margins (about 1 db in PSNR, UCF Sports) in significantly smaller model sizes (84% the size or smaller).

Yiqi Zhong, Zhenyang Ni, Siheng Chen et al.

The historical trajectories previously passing through a location may help infer the future trajectory of an agent currently at this location. Despite great improvements in trajectory forecasting with the guidance of high-definition maps, only a few works have explored such local historical information. In this work, we re-introduce this information as a new type of input data for trajectory forecasting systems: the local behavior data, which we conceptualize as a collection of location-specific historical trajectories. Local behavior data helps the systems emphasize the prediction locality and better understand the impact of static map objects on moving agents. We propose a novel local-behavior-aware (LBA) prediction framework that improves forecasting accuracy by fusing information from observed trajectories, HD maps, and local behavior data. Also, where such historical data is insufficient or unavailable, we employ a local-behavior-free (LBF) prediction framework, which adopts a knowledge-distillation-based architecture to infer the impact of missing data. Extensive experiments demonstrate that upgrading existing methods with these two frameworks significantly improves their performances. Especially, the LBA framework boosts the SOTA methods' performance on the nuScenes dataset by at least 14% for the K=1 metrics.

Cho-Ying Wu, Chin-Cheng Hsu, Ulrich Neumann

This work digs into a root question in human perception: can face geometry be gleaned from one's voices? Previous works that study this question only adopt developments in image synthesis and convert voices into face images to show correlations, but working on the image domain unavoidably involves predicting attributes that voices cannot hint, including facial textures, hairstyles, and backgrounds. We instead investigate the ability to reconstruct 3D faces to concentrate on only geometry, which is much more physiologically grounded. We propose our analysis framework, Cross-Modal Perceptionist, under both supervised and unsupervised learning. First, we construct a dataset, Voxceleb-3D, which extends Voxceleb and includes paired voices and face meshes, making supervised learning possible. Second, we use a knowledge distillation mechanism to study whether face geometry can still be gleaned from voices without paired voices and 3D face data under limited availability of 3D face scans. We break down the core question into four parts and perform visual and numerical analyses as responses to the core question. Our findings echo those in physiology and neuroscience about the correlation between voices and facial structures. The work provides future human-centric cross-modal learning with explainable foundations. See our project page: https://choyingw.github.io/works/Voice2Mesh/index.html

Bohan Tang, Yiqi Zhong, Chenxin Xu et al.

In multi-modal multi-agent trajectory forecasting, two major challenges have not been fully tackled: 1) how to measure the uncertainty brought by the interaction module that causes correlations among the predicted trajectories of multiple agents; 2) how to rank the multiple predictions and select the optimal predicted trajectory. In order to handle these challenges, this work first proposes a novel concept, collaborative uncertainty (CU), which models the uncertainty resulting from interaction modules. Then we build a general CU-aware regression framework with an original permutation-equivariant uncertainty estimator to do both tasks of regression and uncertainty estimation. Further, we apply the proposed framework to current SOTA multi-agent multi-modal forecasting systems as a plugin module, which enables the SOTA systems to 1) estimate the uncertainty in the multi-agent multi-modal trajectory forecasting task; 2) rank the multiple predictions and select the optimal one based on the estimated uncertainty. We conduct extensive experiments on a synthetic dataset and two public large-scale multi-agent trajectory forecasting benchmarks. Experiments show that: 1) on the synthetic dataset, the CU-aware regression framework allows the model to appropriately approximate the ground-truth Laplace distribution; 2) on the multi-agent trajectory forecasting benchmarks, the CU-aware regression framework steadily helps SOTA systems improve their performances. Specially, the proposed framework helps VectorNet improve by 262 cm regarding the Final Displacement Error of the chosen optimal prediction on the nuScenes dataset; 3) for multi-agent multi-modal trajectory forecasting systems, prediction uncertainty is positively correlated with future stochasticity; and 4) the estimated CU values are highly related to the interactive information among agents.

Junying Wang, Yuanlu Xu, Edith Tretschk et al.

The creation of photorealistic dynamic hair remains a major challenge in digital human modeling because of the complex motions, occlusions, and light scattering. Existing methods often resort to static capture and physics-based models that do not scale as they require manual parameter fine-tuning to handle the diversity of hairstyles and motions, and heavy computation to obtain high-quality appearance. In this paper, we present Dynamic Gaussian Hair (DGH), a novel framework that efficiently learns hair dynamics and appearance. We propose: (1) a coarse-to-fine model that learns temporally coherent hair motion dynamics across diverse hairstyles; (2) a strand-guided optimization module that learns a dynamic 3D Gaussian representation for hair appearance with support for differentiable rendering, enabling gradient-based learning of view-consistent appearance under motion. Unlike prior simulation-based pipelines, our approach is fully data-driven, scales with training data, and generalizes across various hairstyles and head motion sequences. Additionally, DGH can be seamlessly integrated into a 3D Gaussian avatar framework, enabling realistic, animatable hair for high-fidelity avatar representation. DGH achieves promising geometry and appearance results, providing a scalable, data-driven alternative to physics-based simulation and rendering.

Cho-Ying Wu, Quankai Gao, Chin-Cheng Hsu et al.

Indoor monocular depth estimation has attracted increasing research interest. Most previous works have been focusing on methodology, primarily experimenting with NYU-Depth-V2 (NYUv2) Dataset, and only concentrated on the overall performance over the test set. However, little is known regarding robustness and generalization when it comes to applying monocular depth estimation methods to real-world scenarios where highly varying and diverse functional \textit{space types} are present such as library or kitchen. A study for performance breakdown into space types is essential to realize a pretrained model's performance variance. To facilitate our investigation for robustness and address limitations of previous works, we collect InSpaceType, a high-quality and high-resolution RGBD dataset for general indoor environments. We benchmark 12 recent methods on InSpaceType and find they severely suffer from performance imbalance concerning space types, which reveals their underlying bias. We extend our analysis to 4 other datasets, 3 mitigation approaches, and the ability to generalize to unseen space types. Our work marks the first in-depth investigation of performance imbalance across space types for indoor monocular depth estimation, drawing attention to potential safety concerns for model deployment without considering space types, and further shedding light on potential ways to improve robustness. See \url{https://depthcomputation.github.io/DepthPublic} for data and the supplementary document. The benchmark list on the GitHub project page keeps updates for the lastest monocular depth estimation methods.

Cho-Ying Wu, Yiqi Zhong, Junying Wang et al.

Indoor robots rely on depth to perform tasks like navigation or obstacle detection, and single-image depth estimation is widely used to assist perception. Most indoor single-image depth prediction focuses less on model generalizability to unseen datasets, concerned with in-the-wild robustness for system deployment. This work leverages gradient-based meta-learning to gain higher generalizability on zero-shot cross-dataset inference. Unlike the most-studied meta-learning of image classification associated with explicit class labels, no explicit task boundaries exist for continuous depth values tied to highly varying indoor environments regarding object arrangement and scene composition. We propose fine-grained task that treats each RGB-D mini-batch as a task in our meta-learning formulation. We first show that our method on limited data induces a much better prior (max 27.8% in RMSE). Then, finetuning on meta-learned initialization consistently outperforms baselines without the meta approach. Aiming at generalization, we propose zero-shot cross-dataset protocols and validate higher generalizability induced by our meta-initialization, as a simple and useful plugin to many existing depth estimation methods. The work at the intersection of depth and meta-learning potentially drives both research to step closer to practical robotic and machine perception usage.

Cho-Ying Wu, Quankai Gao, Chin-Cheng Hsu et al.

Indoor monocular depth estimation helps home automation, including robot navigation or AR/VR for surrounding perception. Most previous methods primarily experiment with the NYUv2 Dataset and concentrate on the overall performance in their evaluation. However, their robustness and generalization to diversely unseen types or categories for indoor spaces (spaces types) have yet to be discovered. Researchers may empirically find degraded performance in a released pretrained model on custom data or less-frequent types. This paper studies the common but easily overlooked factor-space type and realizes a model's performance variances across spaces. We present InSpaceType Dataset, a high-quality RGBD dataset for general indoor scenes, and benchmark 13 recent state-of-the-art methods on InSpaceType. Our examination shows that most of them suffer from performance imbalance between head and tailed types, and some top methods are even more severe. The work reveals and analyzes underlying bias in detail for transparency and robustness. We extend the analysis to a total of 4 datasets and discuss the best practice in synthetic data curation for training indoor monocular depth. Further, dataset ablation is conducted to find out the key factor in generalization. This work marks the first in-depth investigation of performance variances across space types and, more importantly, releases useful tools, including datasets and codes, to closely examine your pretrained depth models. Data and code: https://depthcomputation.github.io/DepthPublic/

Qiangeng Xu, Yiqi Zhong, Ulrich Neumann

Advances in LiDAR sensors provide rich 3D data that supports 3D scene understanding. However, due to occlusion and signal miss, LiDAR point clouds are in practice 2.5D as they cover only partial underlying shapes, which poses a fundamental challenge to 3D perception. To tackle the challenge, we present a novel LiDAR-based 3D object detection model, dubbed Behind the Curtain Detector (BtcDet), which learns the object shape priors and estimates the complete object shapes that are partially occluded (curtained) in point clouds. BtcDet first identifies the regions that are affected by occlusion and signal miss. In these regions, our model predicts the probability of occupancy that indicates if a region contains object shapes. Integrated with this probability map, BtcDet can generate high-quality 3D proposals. Finally, the probability of occupancy is also integrated into a proposal refinement module to generate the final bounding boxes. Extensive experiments on the KITTI Dataset and the Waymo Open Dataset demonstrate the effectiveness of BtcDet. Particularly, for the 3D detection of both cars and cyclists on the KITTI benchmark, BtcDet surpasses all of the published state-of-the-art methods by remarkable margins. Code is released (https://github.com/Xharlie/BtcDet}{https://github.com/Xharlie/BtcDet).

Yiqi Zhong, Cho-Ying Wu, Suya You et al.

In this paper, we propose our Correlation For Completion Network (CFCNet), an end-to-end deep learning model that uses the correlation between two data sources to perform sparse depth completion. CFCNet learns to capture, to the largest extent, the semantically correlated features between RGB and depth information. Through pairs of image pixels and the visible measurements in a sparse depth map, CFCNet facilitates feature-level mutual transformation of different data sources. Such a transformation enables CFCNet to predict features and reconstruct data of missing depth measurements according to their corresponding, transformed RGB features. We extend canonical correlation analysis to a 2D domain and formulate it as one of our training objectives (i.e. 2d deep canonical correlation, or "2D2CCA loss"). Extensive experiments validate the ability and flexibility of our CFCNet compared to the state-of-the-art methods on both indoor and outdoor scenes with different real-life sparse patterns. Codes are available at: https://github.com/choyingw/CFCNet.

Qiangeng Xu, Weiyue Wang, Duygu Ceylan et al.

Reconstructing 3D shapes from single-view images has been a long-standing research problem. In this paper, we present DISN, a Deep Implicit Surface Network which can generate a high-quality detail-rich 3D mesh from an 2D image by predicting the underlying signed distance fields. In addition to utilizing global image features, DISN predicts the projected location for each 3D point on the 2D image, and extracts local features from the image feature maps. Combining global and local features significantly improves the accuracy of the signed distance field prediction, especially for the detail-rich areas. To the best of our knowledge, DISN is the first method that constantly captures details such as holes and thin structures present in 3D shapes from single-view images. DISN achieves the state-of-the-art single-view reconstruction performance on a variety of shape categories reconstructed from both synthetic and real images. Code is available at https://github.com/xharlie/DISN The supplementary can be found at https://xharlie.github.io/images/neurips_2019_supp.pdf

Weiyue Wang, Duygu Ceylan, Radomir Mech et al.

Applications in virtual and augmented reality create a demand for rapid creation and easy access to large sets of 3D models. An effective way to address this demand is to edit or deform existing 3D models based on a reference, e.g., a 2D image which is very easy to acquire. Given such a source 3D model and a target which can be a 2D image, 3D model, or a point cloud acquired as a depth scan, we introduce 3DN, an end-to-end network that deforms the source model to resemble the target. Our method infers per-vertex offset displacements while keeping the mesh connectivity of the source model fixed. We present a training strategy which uses a novel differentiable operation, mesh sampling operator, to generalize our method across source and target models with varying mesh densities. Mesh sampling operator can be seamlessly integrated into the network to handle meshes with different topologies. Qualitative and quantitative results show that our method generates higher quality results compared to the state-of-the art learning-based methods for 3D shape generation. Code is available at github.com/laughtervv/3DN.

Qiangui Huang, Weiyue Wang, Ulrich Neumann

Point clouds are an efficient data format for 3D data. However, existing 3D segmentation methods for point clouds either do not model local dependencies \cite{pointnet} or require added computations \cite{kd-net,pointnet2}. This work presents a novel 3D segmentation framework, RSNet\footnote{Codes are released here https://github.com/qianguih/RSNet}, to efficiently model local structures in point clouds. The key component of the RSNet is a lightweight local dependency module. It is a combination of a novel slice pooling layer, Recurrent Neural Network (RNN) layers, and a slice unpooling layer. The slice pooling layer is designed to project features of unordered points onto an ordered sequence of feature vectors so that traditional end-to-end learning algorithms (RNNs) can be applied. The performance of RSNet is validated by comprehensive experiments on the S3DIS\cite{stanford}, ScanNet\cite{scannet}, and ShapeNet \cite{shapenet} datasets. In its simplest form, RSNets surpass all previous state-of-the-art methods on these benchmarks. And comparisons against previous state-of-the-art methods \cite{pointnet, pointnet2} demonstrate the efficiency of RSNets.

Junying Wang, Jingyuan Liu, Xin Sun et al.

This paper introduces Comprehensive Relighting, the first all-in-one approach that can both control and harmonize the lighting from an image or video of humans with arbitrary body parts from any scene. Building such a generalizable model is extremely challenging due to the lack of dataset, restricting existing image-based relighting models to a specific scenario (e.g., face or static human). To address this challenge, we repurpose a pre-trained diffusion model as a general image prior and jointly model the human relighting and background harmonization in the coarse-to-fine framework. To further enhance the temporal coherence of the relighting, we introduce an unsupervised temporal lighting model that learns the lighting cycle consistency from many real-world videos without any ground truth. In inference time, our temporal lighting module is combined with the diffusion models through the spatio-temporal feature blending algorithms without extra training; and we apply a new guided refinement as a post-processing to preserve the high-frequency details from the input image. In the experiments, Comprehensive Relighting shows a strong generalizability and lighting temporal coherence, outperforming existing image-based human relighting and harmonization methods.

Yiqi Zhong, Luming Liang, Bohan Tang et al.

We introduce motion graph, a novel approach to the video prediction problem, which predicts future video frames from limited past data. The motion graph transforms patches of video frames into interconnected graph nodes, to comprehensively describe the spatial-temporal relationships among them. This representation overcomes the limitations of existing motion representations such as image differences, optical flow, and motion matrix that either fall short in capturing complex motion patterns or suffer from excessive memory consumption. We further present a video prediction pipeline empowered by motion graph, exhibiting substantial performance improvements and cost reductions. Experiments on various datasets, including UCF Sports, KITTI and Cityscapes, highlight the strong representative ability of motion graph. Especially on UCF Sports, our method matches and outperforms the SOTA methods with a significant reduction in model size by 78% and a substantial decrease in GPU memory utilization by 47%.

Quankai Gao, Iliyan Georgiev, Tuanfeng Y. Wang et al.

3D generation has made significant progress, however, it still largely remains at the object-level. Feedforward 3D scene-level generation has been rarely explored due to the lack of models capable of scaling-up latent representation learning on 3D scene-level data. Unlike object-level generative models, which are trained on well-labeled 3D data in a bounded canonical space, scene-level generations with 3D scenes represented by 3D Gaussian Splatting (3DGS) are unbounded and exhibit scale inconsistency across different scenes, making unified latent representation learning for generative purposes extremely challenging. In this paper, we introduce Can3Tok, the first 3D scene-level variational autoencoder (VAE) capable of encoding a large number of Gaussian primitives into a low-dimensional latent embedding, which effectively captures both semantic and spatial information of the inputs. Beyond model design, we propose a general pipeline for 3D scene data processing to address scale inconsistency issue. We validate our method on the recent scene-level 3D dataset DL3DV-10K, where we found that only Can3Tok successfully generalizes to novel 3D scenes, while compared methods fail to converge on even a few hundred scene inputs during training and exhibit zero generalization ability during inference. Finally, we demonstrate image-to-3DGS and text-to-3DGS generation as our applications to demonstrate its ability to facilitate downstream generation tasks.

Jiankun Zhong, Zitong Zhan, Quankai Gao et al.

Structure-from-Motion (SfM), a method that recovers camera poses and scene geometry from uncalibrated images, is a central component in robotic reconstruction and simulation. Despite the state-of-the-art performance of traditional SfM methods such as COLMAP and its follow-up work, GLOMAP, naive CPU-specialized implementations of bundle adjustment (BA) or global positioning (GP) introduce significant computational overhead when handling large-scale scenarios, leading to a trade-off between accuracy and speed in SfM. Moreover, the blessing of efficient C++-based implementations in COLMAP and GLOMAP comes with the curse of limited flexibility, as they lack support for various external optimization options. On the other hand, while deep learning based SfM pipelines like VGGSfM and VGGT enable feed-forward 3D reconstruction, they are unable to scale to thousands of input views at once as GPU memory consumption increases sharply as the number of input views grows. In this paper, we unleash the full potential of GPU parallel computation to accelerate each critical stage of the standard SfM pipeline. Building upon recent advances in sparse-aware bundle adjustment optimization, our design extends these techniques to accelerate both BA and GP within a unified global SfM framework. Through extensive experiments on datasets of varying scales (e.g. 5000 images where VGGSfM and VGGT run out of memory), our method demonstrates up to about 40 times speedup over COLMAP while achieving consistently comparable or even improved reconstruction accuracy. Our project page can be found at https://cre185.github.io/InstantSfM/.

Quankai Gao, Qiangeng Xu, Zhe Cao et al.

Creating 4D fields of Gaussian Splatting from images or videos is a challenging task due to its under-constrained nature. While the optimization can draw photometric reference from the input videos or be regulated by generative models, directly supervising Gaussian motions remains underexplored. In this paper, we introduce a novel concept, Gaussian flow, which connects the dynamics of 3D Gaussians and pixel velocities between consecutive frames. The Gaussian flow can be efficiently obtained by splatting Gaussian dynamics into the image space. This differentiable process enables direct dynamic supervision from optical flow. Our method significantly benefits 4D dynamic content generation and 4D novel view synthesis with Gaussian Splatting, especially for contents with rich motions that are hard to be handled by existing methods. The common color drifting issue that happens in 4D generation is also resolved with improved Guassian dynamics. Superior visual quality on extensive experiments demonstrates our method's effectiveness. Quantitative and qualitative evaluations show that our method achieves state-of-the-art results on both tasks of 4D generation and 4D novel view synthesis. Project page: https://zerg-overmind.github.io/GaussianFlow.github.io/

Cho-Ying Wu, Yiqi Zhong, Junying Wang et al.

Model generalizability to unseen datasets, concerned with in-the-wild robustness, is less studied for indoor single-image depth prediction. We leverage gradient-based meta-learning for higher generalizability on zero-shot cross-dataset inference. Unlike the most-studied image classification in meta-learning, depth is pixel-level continuous range values, and mappings from each image to depth vary widely across environments. Thus no explicit task boundaries exist. We instead propose fine-grained task that treats each RGB-D pair as a task in our meta-optimization. We first show meta-learning on limited data induces much better prior (max +29.4\%). Using meta-learned weights as initialization for following supervised learning, without involving extra data or information, it consistently outperforms baselines without the method. Compared to most indoor-depth methods that only train/ test on a single dataset, we propose zero-shot cross-dataset protocols, closely evaluate robustness, and show consistently higher generalizability and accuracy by our meta-initialization. The work at the intersection of depth and meta-learning potentially drives both research streams to step closer to practical use.

Qiangeng Xu, Zexiang Xu, Julien Philip et al.

Volumetric neural rendering methods like NeRF generate high-quality view synthesis results but are optimized per-scene leading to prohibitive reconstruction time. On the other hand, deep multi-view stereo methods can quickly reconstruct scene geometry via direct network inference. Point-NeRF combines the advantages of these two approaches by using neural 3D point clouds, with associated neural features, to model a radiance field. Point-NeRF can be rendered efficiently by aggregating neural point features near scene surfaces, in a ray marching-based rendering pipeline. Moreover, Point-NeRF can be initialized via direct inference of a pre-trained deep network to produce a neural point cloud; this point cloud can be finetuned to surpass the visual quality of NeRF with 30X faster training time. Point-NeRF can be combined with other 3D reconstruction methods and handles the errors and outliers in such methods via a novel pruning and growing mechanism. The experiments on the DTU, the NeRF Synthetics , the ScanNet and the Tanks and Temples datasets demonstrate Point-NeRF can surpass the existing methods and achieve the state-of-the-art results.

Cho-Ying Wu, Jialiang Wang, Michael Hall et al.

The majority of prior monocular depth estimation methods without groundtruth depth guidance focus on driving scenarios. We show that such methods generalize poorly to unseen complex indoor scenes, where objects are cluttered and arbitrarily arranged in the near field. To obtain more robustness, we propose a structure distillation approach to learn knacks from an off-the-shelf relative depth estimator that produces structured but metric-agnostic depth. By combining structure distillation with a branch that learns metrics from left-right consistency, we attain structured and metric depth for generic indoor scenes and make inferences in real-time. To facilitate learning and evaluation, we collect SimSIN, a dataset from simulation with thousands of environments, and UniSIN, a dataset that contains about 500 real scan sequences of generic indoor environments. We experiment in both sim-to-real and real-to-real settings, and show improvements, as well as in downstream applications using our depth maps. This work provides a full study, covering methods, data, and applications aspects.

Bohan Tang, Yiqi Zhong, Ulrich Neumann et al.

Uncertainty modeling is critical in trajectory forecasting systems for both interpretation and safety reasons. To better predict the future trajectories of multiple agents, recent works have introduced interaction modules to capture interactions among agents. This approach leads to correlations among the predicted trajectories. However, the uncertainty brought by such correlations is neglected. To fill this gap, we propose a novel concept, collaborative uncertainty(CU), which models the uncertainty resulting from the interaction module. We build a general CU-based framework to make a prediction model to learn the future trajectory and the corresponding uncertainty. The CU-based framework is integrated as a plugin module to current state-of-the-art (SOTA) systems and deployed in two special cases based on multivariate Gaussian and Laplace distributions. In each case, we conduct extensive experiments on two synthetic datasets and two public, large-scale benchmarks of trajectory forecasting. The results are promising: 1) The results of synthetic datasets show that CU-based framework allows the model to appropriately approximate the ground-truth distribution. 2) The results of trajectory forecasting benchmarks demonstrate that the CU-based framework steadily helps SOTA systems improve their performances. Especially, the proposed CU-based framework helps VectorNet improve by 57cm regarding Final Displacement Error on nuScenes dataset. 3) The visualization results of CU illustrate that the value of CU is highly related to the amount of the interactive information among agents.

Cho-Ying Wu, Qiangeng Xu, Ulrich Neumann

This work studies learning from a synergy process of 3D Morphable Models (3DMM) and 3D facial landmarks to predict complete 3D facial geometry, including 3D alignment, face orientation, and 3D face modeling. Our synergy process leverages a representation cycle for 3DMM parameters and 3D landmarks. 3D landmarks can be extracted and refined from face meshes built by 3DMM parameters. We next reverse the representation direction and show that predicting 3DMM parameters from sparse 3D landmarks improves the information flow. Together we create a synergy process that utilizes the relation between 3D landmarks and 3DMM parameters, and they collaboratively contribute to better performance. We extensively validate our contribution on full tasks of facial geometry prediction and show our superior and robust performance on these tasks for various scenarios. Particularly, we adopt only simple and widely-used network operations to attain fast and accurate facial geometry prediction. Codes and data: https://choyingw.github.io/works/SynergyNet/

Cho-Ying Wu, Ke Xu, Chin-Cheng Hsu et al.

This work focuses on the analysis that whether 3D face models can be learned from only the speech inputs of speakers. Previous works for cross-modal face synthesis study image generation from voices. However, image synthesis includes variations such as hairstyles, backgrounds, and facial textures, that are arguably irrelevant to voice or without direct studies to show correlations. We instead investigate the ability to reconstruct 3D faces to concentrate on only geometry, which is more physiologically grounded. We propose both the supervised learning and unsupervised learning frameworks. Especially we demonstrate how unsupervised learning is possible in the absence of a direct voice-to-3D-face dataset under limited availability of 3D face scans when the model is equipped with knowledge distillation. To evaluate the performance, we also propose several metrics to measure the geometric fitness of two 3D faces based on points, lines, and regions. We find that 3D face shapes can be reconstructed from voices. Experimental results suggest that 3D faces can be reconstructed from voices, and our method can improve the performance over the baseline. The best performance gains (15% - 20%) on ear-to-ear distance ratio metric (ER) coincides with the intuition that one can roughly envision whether a speaker's face is overall wider or thinner only from a person's voice. See our project page for codes and data.

Cho-Ying Wu, Qiangeng Xu, Ulrich Neumann

This work focuses on complete 3D facial geometry prediction, including 3D facial alignment via 3D face modeling and face orientation estimation using the proposed multi-task, multi-modal, and multi-representation landmark refinement network (M$^3$-LRN). Our focus is on the important facial attributes, 3D landmarks, and we fully utilize their embedded information to guide 3D facial geometry learning. We first propose a multi-modal and multi-representation feature aggregation for landmark refinement. Next, we are the first to study 3DMM regression from sparse 3D landmarks and utilize multi-representation advantage to attain better geometry prediction. We attain the state of the art from extensive experiments on all tasks of learning 3D facial geometry. We closely validate contributions of each modality and representation. Our results are robust across cropped faces, underwater scenarios, and extreme poses. Specially we adopt only simple and widely used network operations in M$^3$-LRN and attain a near 20\% improvement on face orientation estimation over the current best performance. See our project page here.

Cho-Ying Wu, Xiaoyan Hu, Michael Happold et al.

Most previous works of outdoor instance segmentation for images only use color information. We explore a novel direction of sensor fusion to exploit stereo cameras. Geometric information from disparities helps separate overlapping objects of the same or different classes. Moreover, geometric information penalizes region proposals with unlikely 3D shapes thus suppressing false positive detections. Mask regression is based on 2D, 2.5D, and 3D ROI using the pseudo-lidar and image-based representations. These mask predictions are fused by a mask scoring process. However, public datasets only adopt stereo systems with shorter baseline and focal legnth, which limit measuring ranges of stereo cameras. We collect and utilize High-Quality Driving Stereo (HQDS) dataset, using much longer baseline and focal length with higher resolution. Our performance attains state of the art. Please refer to our project page. The full paper is available here.

Cho-Ying Wu, Ulrich Neumann

This paper introduces Scene Completeness-Aware Depth Completion (SCADC) to complete raw lidar scans into dense depth maps with fine and complete scene structures. Recent sparse depth completion for lidars only focuses on the lower scenes and produces irregular estimations on the upper because existing datasets, such as KITTI, do not provide groundtruth for upper areas. These areas are considered less important since they are usually sky or trees of less scene understanding interest. However, we argue that in several driving scenarios such as large trucks or cars with loads, objects could extend to the upper parts of scenes. Thus depth maps with structured upper scene estimation are important for RGBD algorithms. SCADC adopts stereo images that produce disparities with better scene completeness but are generally less precise than lidars, to help sparse lidar depth completion. To our knowledge, we are the first to focus on scene completeness of sparse depth completion. We validate our SCADC on both depth estimate precision and scene-completeness on KITTI. Moreover, we experiment on less-explored outdoor RGBD semantic segmentation with scene completeness-aware D-input to validate our method.

Qiangeng Xu, Xudong Sun, Cho-Ying Wu et al.

Due to the sparsity and irregularity of the point cloud data, methods that directly consume points have become popular. Among all point-based models, graph convolutional networks (GCN) lead to notable performance by fully preserving the data granularity and exploiting point interrelation. However, point-based networks spend a significant amount of time on data structuring (e.g., Farthest Point Sampling (FPS) and neighbor points querying), which limit the speed and scalability. In this paper, we present a method, named Grid-GCN, for fast and scalable point cloud learning. Grid-GCN uses a novel data structuring strategy, Coverage-Aware Grid Query (CAGQ). By leveraging the efficiency of grid space, CAGQ improves spatial coverage while reducing the theoretical time complexity. Compared with popular sampling methods such as Farthest Point Sampling (FPS) and Ball Query, CAGQ achieves up to 50X speed-up. With a Grid Context Aggregation (GCA) module, Grid-GCN achieves state-of-the-art performance on major point cloud classification and segmentation benchmarks with significantly faster runtime than previous studies. Remarkably, Grid-GCN achieves the inference speed of 50fps on ScanNet using 81920 points per scene as input.

Cho-Ying Wu, Ulrich Neumann

In this paper, our goal is salient building outline enhancement and extraction from images taken from consumer cameras using L0 smoothing. We address weak outlines and over-smoothing problem. Weak outlines are often undetected by edge extractors or easily smoothed out. We propose an iterative method, including the smoothing cell and sharpening cell. In the smoothing cell, we iteratively enlarge the smoothing level of the L0 smoothing. In the sharpening cell, we use Hough Transform to extract lines, based on the assumption that salient outlines for buildings are usually straight, and enhance those extracted lines. Our goal is to enhance line structures and do the L0 smoothing simultaneously. Also, we propose to create building masks from semantic segmentation using an encoder-decoder network. The masks filter out irrelevant edges. We also provide an evaluation dataset on this task.

Cho Ying Wu, Ulrich Neumann

In this paper, we propose the multi-domain dictionary learning (MDDL) to make dictionary learning-based classification more robust to data representing in different domains. We use adversarial neural networks to generate data in different styles, and collect all the generated data into a miscellaneous dictionary. To tackle the dictionary learning with many samples, we compute the weighting matrix that compress the miscellaneous dictionary from multi-sample per class to single sample per class. We show that the time complexity solving the proposed MDDL with weighting matrix is the same as solving the dictionary with single sample per class. Moreover, since the weighting matrix could help the solver rely more on the training data, which possibly lie in the same domain with the testing data, the classification could be more accurate.

Qiangeng Xu, Hanwang Zhang, Weiyue Wang et al.

In this paper, we introduce a stochastic dynamics video infilling (SDVI) framework to generate frames between long intervals in a video. Our task differs from video interpolation which aims to produce transitional frames for a short interval between every two frames and increase the temporal resolution. Our task, namely video infilling, however, aims to infill long intervals with plausible frame sequences. Our framework models the infilling as a constrained stochastic generation process and sequentially samples dynamics from the inferred distribution. SDVI consists of two parts: (1) a bi-directional constraint propagation module to guarantee the spatial-temporal coherence among frames, (2) a stochastic sampling process to generate dynamics from the inferred distributions. Experimental results show that SDVI can generate clear frame sequences with varying contents. Moreover, motions in the generated sequence are realistic and able to transfer smoothly from the given start frame to the terminal frame. Our project site is https://xharlie.github.io/projects/project_sites/SDVI/video_results.html

Weiyue Wang, Ulrich Neumann

Convolutional neural networks (CNN) are limited by the lack of capability to handle geometric information due to the fixed grid kernel structure. The availability of depth data enables progress in RGB-D semantic segmentation with CNNs. State-of-the-art methods either use depth as additional images or process spatial information in 3D volumes or point clouds. These methods suffer from high computation and memory cost. To address these issues, we present Depth-aware CNN by introducing two intuitive, flexible and effective operations: depth-aware convolution and depth-aware average pooling. By leveraging depth similarity between pixels in the process of information propagation, geometry is seamlessly incorporated into CNN. Without introducing any additional parameters, both operators can be easily integrated into existing CNNs. Extensive experiments and ablation studies on challenging RGB-D semantic segmentation benchmarks validate the effectiveness and flexibility of our approach.

Qiangui Huang, Kevin Zhou, Suya You et al.

Many state-of-the-art computer vision algorithms use large scale convolutional neural networks (CNNs) as basic building blocks. These CNNs are known for their huge number of parameters, high redundancy in weights, and tremendous computing resource consumptions. This paper presents a learning algorithm to simplify and speed up these CNNs. Specifically, we introduce a "try-and-learn" algorithm to train pruning agents that remove unnecessary CNN filters in a data-driven way. With the help of a novel reward function, our agents removes a significant number of filters in CNNs while maintaining performance at a desired level. Moreover, this method provides an easy control of the tradeoff between network performance and its scale. Per- formance of our algorithm is validated with comprehensive pruning experiments on several popular CNNs for visual recognition and semantic segmentation tasks.

Weiyue Wang, Ronald Yu, Qiangui Huang et al.

We introduce Similarity Group Proposal Network (SGPN), a simple and intuitive deep learning framework for 3D object instance segmentation on point clouds. SGPN uses a single network to predict point grouping proposals and a corresponding semantic class for each proposal, from which we can directly extract instance segmentation results. Important to the effectiveness of SGPN is its novel representation of 3D instance segmentation results in the form of a similarity matrix that indicates the similarity between each pair of points in embedded feature space, thus producing an accurate grouping proposal for each point. To the best of our knowledge, SGPN is the first framework to learn 3D instance-aware semantic segmentation on point clouds. Experimental results on various 3D scenes show the effectiveness of our method on 3D instance segmentation, and we also evaluate the capability of SGPN to improve 3D object detection and semantic segmentation results. We also demonstrate its flexibility by seamlessly incorporating 2D CNN features into the framework to boost performance.

Weiyue Wang, Qiangui Huang, Suya You et al.

Recent advances in convolutional neural networks have shown promising results in 3D shape completion. But due to GPU memory limitations, these methods can only produce low-resolution outputs. To inpaint 3D models with semantic plausibility and contextual details, we introduce a hybrid framework that combines a 3D Encoder-Decoder Generative Adversarial Network (3D-ED-GAN) and a Long-term Recurrent Convolutional Network (LRCN). The 3D-ED-GAN is a 3D convolutional neural network trained with a generative adversarial paradigm to fill missing 3D data in low-resolution. LRCN adopts a recurrent neural network architecture to minimize GPU memory usage and incorporates an Encoder-Decoder pair into a Long Short-term Memory Network. By handling the 3D model as a sequence of 2D slices, LRCN transforms a coarse 3D shape into a more complete and higher resolution volume. While 3D-ED-GAN captures global contextual structure of the 3D shape, LRCN localizes the fine-grained details. Experimental results on both real-world and synthetic data show reconstructions from corrupted models result in complete and high-resolution 3D objects.

Qiangui Huang, Weiyue Wang, Kevin Zhou et al.

Recurrent neural network (RNN), as a powerful contextual dependency modeling framework, has been widely applied to scene labeling problems. However, this work shows that directly applying traditional RNN architectures, which unfolds a 2D lattice grid into a sequence, is not sufficient to model structure dependencies in images due to the "impact vanishing" problem. First, we give an empirical analysis about the "impact vanishing" problem. Then, a new RNN unit named Recurrent Neural Network with explicit long range conditioning (RNN-ELC) is designed to alleviate this problem. A novel neural network architecture is built for scene labeling tasks where one of the variants of the new RNN unit, Gated Recurrent Unit with Explicit Long-range Conditioning (GRU-ELC), is used to model multi scale contextual dependencies in images. We validate the use of GRU-ELC units with state-of-the-art performance on three standard scene labeling datasets. Comprehensive experiments demonstrate that the new GRU-ELC unit benefits scene labeling problem a lot as it can encode longer contextual dependencies in images more effectively than traditional RNN units.