We present a novel method for simultaneous learning of depth, egomotion, object motion, and camera intrinsics from monocular videos, using only consistency across neighboring video frames as supervision signal. Similarly to prior work, our method learns by applying differentiable warping to frames and comparing the result to adjacent ones, but it provides several improvements: We address occlusions geometrically and differentiably, directly using the depth maps as predicted during training. We introduce randomized layer normalization, a novel powerful regularizer, and we account for object motion relative to the scene. To the best of our knowledge, our work is the first to learn the camera intrinsic parameters, including lens distortion, from video in an unsupervised manner, thereby allowing us to extract accurate depth and motion from arbitrary videos of unknown origin at scale. We evaluate our results on the Cityscapes, KITTI and EuRoC datasets, establishing new state of the art on depth prediction and odometry, and demonstrate qualitatively that depth prediction can be learned from a collection of YouTube videos.
Recovering the shape and reflectance of non-Lambertian surfaces remains a challenging problem in computer vision since the view-dependent appearance invalidates traditional photo-consistency constraint. In this paper, we introduce a novel concentric multi-spectral light field (CMSLF) design that is able to recover the shape and reflectance of surfaces with arbitrary material in one shot. Our CMSLF system consists of an array of cameras arranged on concentric circles where each ring captures a specific spectrum. Coupled with a multi-spectral ring light, we are able to sample viewpoint and lighting variations in a single shot via spectral multiplexing. We further show that such concentric camera/light setting results in a unique pattern of specular changes across views that enables robust depth estimation. We formulate a physical-based reflectance model on CMSLF to estimate depth and multi-spectral reflectance map without imposing any surface prior. Extensive synthetic and real experiments show that our method outperforms state-of-the-art light field-based techniques, especially in non-Lambertian scenes.
We propose a novel plug-and-play (PnP) module for improving depth prediction with taking arbitrary patterns of sparse depths as input. Given any pre-trained depth prediction model, our PnP module updates the intermediate feature map such that the model outputs new depths consistent with the given sparse depths. Our method requires no additional training and can be applied to practical applications such as leveraging both RGB and sparse LiDAR points to robustly estimate dense depth map. Our approach achieves consistent improvements on various state-of-the-art methods on indoor (i.e., NYU-v2) and outdoor (i.e., KITTI) datasets. Various types of LiDARs are also synthesized in our experiments to verify the general applicability of our PnP module in practice. For project page, see https://zswang666.github.io/PnP-Depth-Project-Page/
Recent works have proven that many relevant visual tasks are closely related one to another. Yet, this connection is seldom deployed in practice due to the lack of practical methodologies to transfer learned concepts across different training processes. In this work, we introduce a novel adaptation framework that can operate across both task and domains. Our framework learns to transfer knowledge across tasks in a fully supervised domain (e.g., synthetic data) and use this knowledge on a different domain where we have only partial supervision (e.g., real data). Our proposal is complementary to existing domain adaptation techniques and extends them to cross tasks scenarios providing additional performance gains. We prove the effectiveness of our framework across two challenging tasks (i.e., monocular depth estimation and semantic segmentation) and four different domains (Synthia, Carla, Kitti, and Cityscapes).
Recently, convolutional neural networks (CNNs) have shown great success on the task of monocular depth estimation. A fundamental yet unanswered question is: how CNNs can infer depth from a single image. Toward answering this question, we consider visualization of inference of a CNN by identifying relevant pixels of an input image to depth estimation. We formulate it as an optimization problem of identifying the smallest number of image pixels from which the CNN can estimate a depth map with the minimum difference from the estimate from the entire image. To cope with a difficulty with optimization through a deep CNN, we propose to use another network to predict those relevant image pixels in a forward computation. In our experiments, we first show the effectiveness of this approach, and then apply it to different depth estimation networks on indoor and outdoor scene datasets. The results provide several findings that help exploration of the above question.
Depth estimation from a single image represents a fascinating, yet challenging problem with countless applications. Recent works proved that this task could be learned without direct supervision from ground truth labels leveraging image synthesis on sequences or stereo pairs. Focusing on this second case, in this paper we leverage stereo matching in order to improve monocular depth estimation. To this aim we propose monoResMatch, a novel deep architecture designed to infer depth from a single input image by synthesizing features from a different point of view, horizontally aligned with the input image, performing stereo matching between the two cues. In contrast to previous works sharing this rationale, our network is the first trained end-to-end from scratch. Moreover, we show how obtaining proxy ground truth annotation through traditional stereo algorithms, such as Semi-Global Matching, enables more accurate monocular depth estimation still countering the need for expensive depth labels by keeping a self-supervised approach. Exhaustive experimental results prove how the synergy between i) the proposed monoResMatch architecture and ii) proxy-supervision attains state-of-the-art for self-supervised monocular depth estimation. The code is publicly available at https://github.com/fabiotosi92/monoResMatch-Tensorflow.
Real world applications of stereo depth estimation require models that are robust to dynamic variations in the environment. Even though deep learning based stereo methods are successful, they often fail to generalize to unseen variations in the environment, making them less suitable for practical applications such as autonomous driving. In this work, we introduce a "learning-to-adapt" framework that enables deep stereo methods to continuously adapt to new target domains in an unsupervised manner. Specifically, our approach incorporates the adaptation procedure into the learning objective to obtain a base set of parameters that are better suited for unsupervised online adaptation. To further improve the quality of the adaptation, we learn a confidence measure that effectively masks the errors introduced during the unsupervised adaptation. We evaluate our method on synthetic and real-world stereo datasets and our experiments evidence that learning-to-adapt is, indeed beneficial for online adaptation on vastly different domains.
Single-view depth estimation suffers from the problem that a network trained on images from one camera does not generalize to images taken with a different camera model. Thus, changing the camera model requires collecting an entirely new training dataset. In this work, we propose a new type of convolution that can take the camera parameters into account, thus allowing neural networks to learn calibration-aware patterns. Experiments confirm that this improves the generalization capabilities of depth prediction networks considerably, and clearly outperforms the state of the art when the train and test images are acquired with different cameras.
Supervised depth estimation has achieved high accuracy due to the advanced deep network architectures. Since the groundtruth depth labels are hard to obtain, recent methods try to learn depth estimation networks in an unsupervised way by exploring unsupervised cues, which are effective but less reliable than true labels. An emerging way to resolve this dilemma is to transfer knowledge from synthetic images with ground truth depth via domain adaptation techniques. However, these approaches overlook specific geometric structure of the natural images in the target domain (i.e., real data), which is important for high-performing depth prediction. Motivated by the observation, we propose a geometry-aware symmetric domain adaptation framework (GASDA) to explore the labels in the synthetic data and epipolar geometry in the real data jointly. Moreover, by training two image style translators and depth estimators symmetrically in an end-to-end network, our model achieves better image style transfer and generates high-quality depth maps. The experimental results demonstrate the effectiveness of our proposed method and comparable performance against the state-of-the-art. Code will be publicly available at: https://github.com/sshan-zhao/GASDA.
Three-dimensional (3D) reconstruction and scene depth estimation from 2-dimensional (2D) images are major tasks in computer vision. However, using conventional 3D reconstruction techniques gets challenging in participating media such as murky water, fog, or smoke. We have developed a method that uses a time-of-flight (ToF) camera to estimate an object region and depth in participating media simultaneously. The scattering component is saturated, so it does not depend on the scene depth, and received signals bouncing off distant points are negligible due to light attenuation in the participating media, so the observation of such a point contains only a scattering component. These phenomena enable us to estimate the scattering component in an object region from a background that only contains the scattering component. The problem is formulated as robust estimation where the object region is regarded as outliers, and it enables the simultaneous estimation of an object region and depth on the basis of an iteratively reweighted least squares (IRLS) optimization scheme. We demonstrate the effectiveness of the proposed method using captured images from a Kinect v2 in real foggy scenes and evaluate the applicability with synthesized data.
Synthesizing a densely sampled light field from a single image is highly beneficial for many applications. The conventional method reconstructs a depth map and relies on physical-based rendering and a secondary network to improve the synthesized novel views. Simple pixel-based loss also limits the network by making it rely on pixel intensity cue rather than geometric reasoning. In this study, we show that a different geometric representation, namely, appearance flow, can be used to synthesize a light field from a single image robustly and directly. A single end-to-end deep neural network that does not require a physical-based approach nor a post-processing subnetwork is proposed. Two novel loss functions based on known light field domain knowledge are presented to enable the network to preserve the spatio-angular consistency between sub-aperture images effectively. Experimental results show that the proposed model successfully synthesizes dense light fields and qualitatively and quantitatively outperforms the previous model . The method can be generalized to arbitrary scenes, rather than focusing on a particular class of object. The synthesized light field can be used for various applications, such as depth estimation and refocusing.
Robust geometric and semantic scene understanding is ever more important in many real-world applications such as autonomous driving and robotic navigation. In this paper, we propose a multi-task learning-based approach capable of jointly performing geometric and semantic scene understanding, namely depth prediction (monocular depth estimation and depth completion) and semantic scene segmentation. Within a single temporally constrained recurrent network, our approach uniquely takes advantage of a complex series of skip connections, adversarial training and the temporal constraint of sequential frame recurrence to produce consistent depth and semantic class labels simultaneously. Extensive experimental evaluation demonstrates the efficacy of our approach compared to other contemporary state-of-the-art techniques.
Monocular 3D scene understanding tasks, such as object size estimation, heading angle estimation and 3D localization, is challenging. Successful modern day methods for 3D scene understanding require the use of a 3D sensor. On the other hand, single image based methods have significantly worse performance. In this work, we aim at bridging the performance gap between 3D sensing and 2D sensing for 3D object detection by enhancing LiDAR-based algorithms to work with single image input. Specifically, we perform monocular depth estimation and lift the input image to a point cloud representation, which we call pseudo-LiDAR point cloud. Then we can train a LiDAR-based 3D detection network with our pseudo-LiDAR end-to-end. Following the pipeline of two-stage 3D detection algorithms, we detect 2D object proposals in the input image and extract a point cloud frustum from the pseudo-LiDAR for each proposal. Then an oriented 3D bounding box is detected for each frustum. To handle the large amount of noise in the pseudo-LiDAR, we propose two innovations: (1) use a 2D-3D bounding box consistency constraint, adjusting the predicted 3D bounding box to have a high overlap with its corresponding 2D proposal after projecting onto the image; (2) use the instance mask instead of the bounding box as the representation of 2D proposals, in order to reduce the number of points not belonging to the object in the point cloud frustum. Through our evaluation on the KITTI benchmark, we achieve the top-ranked performance on both bird's eye view and 3D object detection among all monocular methods, effectively quadrupling the performance over previous state-of-the-art. Our code is available at https://github.com/xinshuoweng/Mono3D_PLiDAR.
In this paper, we proposed a new deep learning based dense monocular SLAM method. Compared to existing methods, the proposed framework constructs a dense 3D model via a sparse to dense mapping using learned surface normals. With single view learned depth estimation as prior for monocular visual odometry, we obtain both accurate positioning and high quality depth reconstruction. The depth and normal are predicted by a single network trained in a tightly coupled manner.Experimental results show that our method significantly improves the performance of visual tracking and depth prediction in comparison to the state-of-the-art in deep monocular dense SLAM.
Convolutional neural networks (CNN) have shown state-of-the-art results for low-level computer vision problems such as stereo and monocular disparity estimations, but still, have much room to further improve their performance in terms of accuracy, numbers of parameters, etc. Recent works have uncovered the advantages of using an unsupervised scheme to train CNN's to estimate monocular disparity, where only the relatively-easy-to-obtain stereo images are needed for training. We propose a novel encoder-decoder architecture that outperforms previous unsupervised monocular depth estimation networks by (i) taking into account ambiguities, (ii) efficient fusion between encoder and decoder features with rectangular convolutions and (iii) domain transformations between encoder and decoder. Our architecture outperforms the Monodepth baseline in all metrics, even with a considerable reduction of parameters. Furthermore, our architecture is capable of estimating a full disparity map in a single forward pass, whereas the baseline needs two passes. We perform extensive experiments to verify the effectiveness of our method on the KITTI dataset.
Supervised learning methods to infer (hypothesize) depth of a scene from a single image require costly per-pixel ground-truth. We follow a geometric approach that exploits abundant stereo imagery to learn a model to hypothesize scene structure without direct supervision. Although we train a network with stereo pairs, we only require a single image at test time to hypothesize disparity or depth. We propose a novel objective function that exploits the bilateral cyclic relationship between the left and right disparities and we introduce an adaptive regularization scheme that allows the network to handle both the co-visible and occluded regions in a stereo pair. This process ultimately produces a model to generate hypotheses for the 3-dimensional structure of the scene as viewed in a single image. When used to generate a single (most probable) estimate of depth, our method outperforms state-of-the-art unsupervised monocular depth prediction methods on the KITTI benchmarks. We show that our method generalizes well by applying our models trained on KITTI to the Make3d dataset.
Depth estimation is an important capability for autonomous vehicles to understand and reconstruct 3D environments as well as avoid obstacles during the execution. Accurate depth sensors such as LiDARs are often heavy, expensive and can only provide sparse depth while lighter depth sensors such as stereo cameras are noiser in comparison. We propose an end-to-end learning algorithm that is capable of using sparse, noisy input depth for refinement and depth completion. Our model also produces the camera pose as a byproduct, making it a great solution for autonomous systems. We evaluate our approach on both indoor and outdoor datasets. Empirical results show that our method performs well on the KITTI~\cite{kitti_geiger2012we} dataset when compared to other competing methods, while having superior performance in dealing with sparse, noisy input depth on the TUM~\cite{sturm12iros} dataset.
Monocular Odometry systems can be broadly categorized as being either Direct, Indirect, or a hybrid of both. While Indirect systems process an alternative image representation to compute geometric residuals, Direct methods process the image pixels directly to generate photometric residuals. Both paradigms have distinct but often complementary properties. This paper presents a Unified Formulation for Visual Odometry, referred to as UFVO, with the following key contributions: (1) a tight coupling of photometric (Direct) and geometric (Indirect) measurements using a joint multi-objective optimization, (2) the use of a utility function as a decision maker that incorporates prior knowledge on both paradigms, (3) descriptor sharing, where a feature can have more than one type of descriptor and its different descriptors are used for tracking and mapping, (4) the depth estimation of both corner features and pixel features within the same map using an inverse depth parametrization, and (5) a corner and pixel selection strategy that extracts both types of information, while promoting a uniform distribution over the image domain. Experiments show that our proposed system can handle large inter-frame motions, inherits the sub-pixel accuracy of direct methods, can run efficiently in real-time, can generate an Indirect map representation at a marginal computational cost when compared to traditional Indirect systems, all while outperforming state of the art in Direct, Indirect and hybrid systems.
Nowadays, the majority of state of the art monocular depth estimation techniques are based on supervised deep learning models. However, collecting RGB images with associated depth maps is a very time consuming procedure. Therefore, recent works have proposed deep architectures for addressing the monocular depth prediction task as a reconstruction problem, thus avoiding the need of collecting ground-truth depth. Following these works, we propose a novel self-supervised deep model for estimating depth maps. Our framework exploits two main strategies: refinement via cycle-inconsistency and distillation. Specifically, first a \emph{student} network is trained to predict a disparity map such as to recover from a frame in a camera view the associated image in the opposite view. Then, a backward cycle network is applied to the generated image to re-synthesize back the input image, estimating the opposite disparity. A third network exploits the inconsistency between the original and the reconstructed input frame in order to output a refined depth map. Finally, knowledge distillation is exploited, such as to transfer information from the refinement network to the student. Our extensive experimental evaluation demonstrate the effectiveness of the proposed framework which outperforms state of the art unsupervised methods on the KITTI benchmark.
In this work, we consider transferring the structure information from large networks to compact ones for dense prediction tasks in computer vision. Previous knowledge distillation strategies used for dense prediction tasks often directly borrow the distillation scheme for image classification and perform knowledge distillation for each pixel separately, leading to sub-optimal performance. Here we propose to distill structured knowledge from large networks to compact networks, taking into account the fact that dense prediction is a structured prediction problem. Specifically, we study two structured distillation schemes: i) pair-wise distillation that distills the pair-wise similarities by building a static graph; and ii) holistic distillation that uses adversarial training to distill holistic knowledge. The effectiveness of our knowledge distillation approaches is demonstrated by experiments on three dense prediction tasks: semantic segmentation, depth estimation and object detection. Code is available at: https://git.io/StructKD