3D object detection from monocular images is an ill-posed problem due to the projective entanglement of depth and scale. To overcome this ambiguity, we present a novel self-supervised method for textured 3D shape reconstruction and pose estimation of rigid objects with the help of strong shape priors and 2D instance masks. Our method predicts the 3D location and meshes of each object in an image using differentiable rendering and a self-supervised objective derived from a pretrained monocular depth estimation network. We use the KITTI 3D object detection dataset to evaluate the accuracy of the method. Experiments demonstrate that we can effectively use noisy monocular depth and differentiable rendering as an alternative to expensive 3D ground-truth labels or LiDAR information.
Self-supervised monocular depth estimation has become an appealing solution to the lack of ground truth labels, but its reconstruction loss often produces over-smoothed results across object boundaries and is incapable of handling occlusion explicitly. In this paper, we propose a new approach to leverage pseudo ground truth depth maps of stereo images generated from self-supervised stereo matching methods. The confidence map of the pseudo ground truth depth map is estimated to mitigate performance degeneration by inaccurate pseudo depth maps. To cope with the prediction error of the confidence map itself, we also leverage the threshold network that learns the threshold dynamically conditioned on the pseudo depth maps. The pseudo depth labels filtered out by the thresholded confidence map are used to supervise the monocular depth network. Furthermore, we propose the probabilistic framework that refines the monocular depth map with the help of its uncertainty map through the pixel-adaptive convolution (PAC) layer. Experimental results demonstrate superior performance to state-of-the-art monocular depth estimation methods. Lastly, we exhibit that the proposed threshold learning can also be used to improve the performance of existing confidence estimation approaches.
Single-view depth estimation (SVDE) plays a crucial role in scene understanding for AR applications, 3D modeling, and robotics, providing the geometry of a scene based on a single image. Recent works have shown that a successful solution strongly relies on the diversity and volume of training data. This data can be sourced from stereo movies and photos. However, they do not provide geometrically complete depth maps (as disparities contain unknown shift value). Therefore, existing models trained on this data are not able to recover correct 3D representations. Our work shows that a model trained on this data along with conventional datasets can gain accuracy while predicting correct scene geometry. Surprisingly, only a small portion of geometrically correct depth maps are required to train a model that performs equally to a model trained on the full geometrically correct dataset. After that, we train computationally efficient models on a mixture of datasets using the proposed method. Through quantitative comparison on completely unseen datasets and qualitative comparison of 3D point clouds, we show that our model defines the new state of the art in general-purpose SVDE.
Depth estimation from monocular images is a challenging problem in computer vision. In this paper, we tackle this problem using a novel network architecture using multi scale feature fusion. Our network uses two different blocks, first which uses different filter sizes for convolution and merges all the individual feature maps. The second block uses dilated convolutions in place of fully connected layers thus reducing computations and increasing the receptive field. We present a new loss function for training the network which uses a depth regression term, SSIM loss term and a multinomial logistic loss term combined. We train and test our network on Make 3D dataset, NYU Depth V2 dataset and Kitti dataset using standard evaluation metrics for depth estimation comprised of RMSE loss and SILog loss. Our network outperforms previous state of the art methods with lesser parameters.
In the recent years, many methods demonstrated the ability of neural networks to learn depth and pose changes in a sequence of images, using only self-supervision as the training signal. Whilst the networks achieve good performance, the often over-looked detail is that due to the inherent ambiguity of monocular vision they predict depth up to an unknown scaling factor. The scaling factor is then typically obtained from the LiDAR ground truth at test time, which severely limits practical applications of these methods. In this paper, we show that incorporating prior information about the camera configuration and the environment, we can remove the scale ambiguity and predict depth directly, still using the self-supervised formulation and not relying on any additional sensors.
It is a classical compute vision problem to obtain real scene depth maps by using a monocular camera, which has been widely concerned in recent years. However, training this model usually requires a large number of artificially labeled samples. To solve this problem, some researchers use a self-supervised learning model to overcome this problem and reduce the dependence on manually labeled data. Nevertheless, the accuracy and reliability of these methods have not reached the expected standard. In this paper, we propose a new self-supervised learning method based on cascade networks. Compared with the previous self-supervised methods, our method has improved accuracy and reliability, and we have proved this by experiments. We show a cascaded neural network that divides the target scene into parts of different sight distances and trains them separately to generate a better depth map. Our approach is divided into the following four steps. In the first step, we use the self-supervised model to estimate the depth of the scene roughly. In the second step, the depth of the scene generated in the first step is used as a label to divide the scene into different depth parts. The third step is to use models with different parameters to generate depth maps of different depth parts in the target scene, and the fourth step is to fuse the depth map. Through the ablation study, we demonstrated the effectiveness of each component individually and showed high-quality, state-of-the-art results in the KITTI benchmark.
Stereo vision generally involves the computation of pixel correspondences and estimation of disparities between rectified image pairs. In many applications, including simultaneous localization and mapping (SLAM) and 3D object detection, the disparities are primarily needed to calculate depth values and the accuracy of depth estimation is often more compelling than disparity estimation. The accuracy of disparity estimation, however, does not directly translate to the accuracy of depth estimation, especially for faraway objects. In the context of learning-based stereo systems, this is largely due to biases imposed by the choices of the disparity-based loss function and the training data. Consequently, the learning algorithms often produce unreliable depth estimates of foreground objects, particularly at large distances~($>50$m). To resolve this issue, we first analyze the effect of those biases and then propose a pair of novel depth-based loss functions for foreground and background, separately. These loss functions are tunable and can balance the inherent bias of the stereo learning algorithms. The efficacy of our solution is demonstrated by an extensive set of experiments, which are benchmarked against state of the art. We show on KITTI~2015 benchmark that our proposed solution yields substantial improvements in disparity and depth estimation, particularly for objects located at distances beyond 50 meters, outperforming the previous state of the art by $10\%$.
We propose a method to compute depth maps for every sub-aperture image in a light field in a view consistent way. Previous light field depth estimation methods typically estimate a depth map only for the central sub-aperture view, and struggle with view consistent estimation. Our method precisely defines depth edges via EPIs, then we diffuse these edges spatially within the central view. These depth estimates are then propagated to all other views in an occlusion-aware way. Finally, disoccluded regions are completed by diffusion in EPI space. Our method runs efficiently with respect to both other classical and deep learning-based approaches, and achieves competitive quantitative metrics and qualitative performance on both synthetic and real-world light fields
Rain fills the atmosphere with water particles, which breaks the common assumption that light travels unaltered from the scene to the camera. While it is well-known that rain affects computer vision algorithms, quantifying its impact is difficult. In this context, we present a rain rendering pipeline that enables the systematic evaluation of common computer vision algorithms to controlled amounts of rain. We present three different ways to add synthetic rain to existing images datasets: completely physic-based; completely data-driven; and a combination of both. The physic-based rain augmentation combines a physical particle simulator and accurate rain photometric modeling. We validate our rendering methods with a user study, demonstrating our rain is judged as much as 73% more realistic than the state-of-theart. Using our generated rain-augmented KITTI, Cityscapes, and nuScenes datasets, we conduct a thorough evaluation of object detection, semantic segmentation, and depth estimation algorithms and show that their performance decreases in degraded weather, on the order of 15% for object detection, 60% for semantic segmentation, and 6-fold increase in depth estimation error. Finetuning on our augmented synthetic data results in improvements of 21% on object detection, 37% on semantic segmentation, and 8% on depth estimation.
Visual odometry (VO) is a prevalent way to deal with the relative localization problem, which is becoming increasingly mature and accurate, but it tends to be fragile under challenging environments. Comparing with classical geometry-based methods, deep learning-based methods can automatically learn effective and robust representations, such as depth, optical flow, feature, ego-motion, etc., from data without explicit computation. Nevertheless, there still lacks a thorough review of the recent advances of deep learning-based VO (Deep VO). Therefore, this paper aims to gain a deep insight on how deep learning can profit and optimize the VO systems. We first screen out a number of qualifications including accuracy, efficiency, scalability, dynamicity, practicability, and extensibility, and employ them as the criteria. Then, using the offered criteria as the uniform measurements, we detailedly evaluate and discuss how deep learning improves the performance of VO from the aspects of depth estimation, feature extraction and matching, pose estimation. We also summarize the complicated and emerging areas of Deep VO, such as mobile robots, medical robots, augmented reality and virtual reality, etc. Through the literature decomposition, analysis, and comparison, we finally put forward a number of open issues and raise some future research directions in this field.
Modern high-definition LIDAR is expensive for commercial autonomous driving vehicles and small indoor robots. An affordable solution to this problem is fusion of planar LIDAR with RGB images to provide a similar level of perception capability. Even though state-of-the-art methods provide approaches to predict depth information from limited sensor input, they are usually a simple concatenation of sparse LIDAR features and dense RGB features through an end-to-end fusion architecture. In this paper, we introduce an inductive late-fusion block which better fuses different sensor modalities inspired by a probability model. The proposed demonstration and aggregation network propagates the mixed context and depth features to the prediction network and serves as a prior knowledge of the depth completion. This late-fusion block uses the dense context features to guide the depth prediction based on demonstrations by sparse depth features. In addition to evaluating the proposed method on benchmark depth completion datasets including NYUDepthV2 and KITTI, we also test the proposed method on a simulated planar LIDAR dataset. Our method shows promising results compared to previous approaches on both the benchmark datasets and simulated dataset with various 3D densities.
Many interesting tasks in machine learning and computer vision are learned by optimising an objective function defined as a weighted linear combination of multiple losses. The final performance is sensitive to choosing the correct (relative) weights for these losses. Finding a good set of weights is often done by adopting them into the set of hyper-parameters, which are set using an extensive grid search. This is computationally expensive. In this paper, we propose a weighting scheme based on the coefficient of variations and set the weights based on properties observed while training the model. The proposed method incorporates a measure of uncertainty to balance the losses, and as a result the loss weights evolve during training without requiring another (learning based) optimisation. In contrast to many loss weighting methods in literature, we focus on single-task multi-loss problems, such as monocular depth estimation and semantic segmentation, and show that multi-task approaches for loss weighting do not work on those single-tasks. The validity of the approach is shown empirically for depth estimation and semantic segmentation on multiple datasets.
Accurate real depth annotations are difficult to acquire, needing the use of special devices such as a LiDAR sensor. Self-supervised methods try to overcome this problem by processing video or stereo sequences, which may not always be available. Instead, in this paper, we propose a domain adaptation approach to train a monocular depth estimation model using a fully-annotated source dataset and a non-annotated target dataset. We bridge the domain gap by leveraging semantic predictions and low-level edge features to provide guidance for the target domain. We enforce consistency between the main model and a second model trained with semantic segmentation and edge maps, and introduce priors in the form of instance heights. Our approach is evaluated on standard domain adaptation benchmarks for monocular depth estimation and show consistent improvement upon the state-of-the-art.
In this paper, we propose a Bidirectional Attention Network (BANet), an end-to-end framework for monocular depth estimation (MDE) that addresses the limitation of effectively integrating local and global information in convolutional neural networks. The structure of this mechanism derives from a strong conceptual foundation of neural machine translation, and presents a light-weight mechanism for adaptive control of computation similar to the dynamic nature of recurrent neural networks. We introduce bidirectional attention modules that utilize the feed-forward feature maps and incorporate the global context to filter out ambiguity. Extensive experiments reveal the high degree of capability of this bidirectional attention model over feed-forward baselines and other state-of-the-art methods for monocular depth estimation on two challenging datasets -- KITTI and DIODE. We show that our proposed approach either outperforms or performs at least on a par with the state-of-the-art monocular depth estimation methods with less memory and computational complexity.
Current capsule endoscopes and next-generation robotic capsules for diagnosis and treatment of gastrointestinal diseases are complex cyber-physical platforms that must orchestrate complex software and hardware functions. The desired tasks for these systems include visual localization, depth estimation, 3D mapping, disease detection and segmentation, automated navigation, active control, path realization and optional therapeutic modules such as targeted drug delivery and biopsy sampling. Data-driven algorithms promise to enable many advanced functionalities for capsule endoscopes, but real-world data is challenging to obtain. Physically-realistic simulations providing synthetic data have emerged as a solution to the development of data-driven algorithms. In this work, we present a comprehensive simulation platform for capsule endoscopy operations and introduce VR-Caps, a virtual active capsule environment that simulates a range of normal and abnormal tissue conditions (e.g., inflated, dry, wet etc.) and varied organ types, capsule endoscope designs (e.g., mono, stereo, dual and 360{\deg}camera), and the type, number, strength, and placement of internal and external magnetic sources that enable active locomotion. VR-Caps makes it possible to both independently or jointly develop, optimize, and test medical imaging and analysis software for the current and next-generation endoscopic capsule systems. To validate this approach, we train state-of-the-art deep neural networks to accomplish various medical image analysis tasks using simulated data from VR-Caps and evaluate the performance of these models on real medical data. Results demonstrate the usefulness and effectiveness of the proposed virtual platform in developing algorithms that quantify fractional coverage, camera trajectory, 3D map reconstruction, and disease classification.
3D photography is a new medium that allows viewers to more fully experience a captured moment. In this work, we refer to a 3D photo as one that displays parallax induced by moving the viewpoint (as opposed to a stereo pair with a fixed viewpoint). 3D photos are static in time, like traditional photos, but are displayed with interactive parallax on mobile or desktop screens, as well as on Virtual Reality devices, where viewing it also includes stereo. We present an end-to-end system for creating and viewing 3D photos, and the algorithmic and design choices therein. Our 3D photos are captured in a single shot and processed directly on a mobile device. The method starts by estimating depth from the 2D input image using a new monocular depth estimation network that is optimized for mobile devices. It performs competitively to the state-of-the-art, but has lower latency and peak memory consumption and uses an order of magnitude fewer parameters. The resulting depth is lifted to a layered depth image, and new geometry is synthesized in parallax regions. We synthesize color texture and structures in the parallax regions as well, using an inpainting network, also optimized for mobile devices, on the LDI directly. Finally, we convert the result into a mesh-based representation that can be efficiently transmitted and rendered even on low-end devices and over poor network connections. Altogether, the processing takes just a few seconds on a mobile device, and the result can be instantly viewed and shared. We perform extensive quantitative evaluation to validate our system and compare its new components against the current state-of-the-art.
Good local features improve the robustness of many 3D re-localization and multi-view reconstruction pipelines. The problem is that viewing angle and distance severely impact the recognizability of a local feature. Attempts to improve appearance invariance by choosing better local feature points or by leveraging outside information, have come with pre-requisites that made some of them impractical. In this paper, we propose a surprisingly effective enhancement to local feature extraction, which improves matching. We show that CNN-based depths inferred from single RGB images are quite helpful, despite their flaws. They allow us to pre-warp images and rectify perspective distortions, to significantly enhance SIFT and BRISK features, enabling more good matches, even when cameras are looking at the same scene but in opposite directions.
Current computer vision tasks based on deep learning require a huge amount of data with annotations for model training or testing, especially in some dense estimation tasks, such as optical flow segmentation and depth estimation. In practice, manual labeling for dense estimation tasks is very difficult or even impossible, and the scenes of the dataset are often restricted to a small range, which dramatically limits the development of the community. To overcome this deficiency, we propose a synthetic dataset generation method to obtain the expandable dataset without burdensome manual workforce. By this method, we construct a dataset called MineNavi containing video footages from first-perspective-view of the aircraft matched with accurate ground truth for depth estimation in aircraft navigation application. We also provide quantitative experiments to prove that pre-training via our MineNavi dataset can improve the performance of depth estimation model and speed up the convergence of the model on real scene data. Since the synthetic dataset has a similar effect to the real-world dataset in the training process of deep model, we also provide additional experiments with monocular depth estimation method to demonstrate the impact of various factors in our dataset such as lighting conditions and motion mode.
Learning-based multi-view stereo (MVS) methods have demonstrated promising results. However, very few existing networks explicitly take the pixel-wise visibility into consideration, resulting in erroneous cost aggregation from occluded pixels. In this paper, we explicitly infer and integrate the pixel-wise occlusion information in the MVS network via the matching uncertainty estimation. The pair-wise uncertainty map is jointly inferred with the pair-wise depth map, which is further used as weighting guidance during the multi-view cost volume fusion. As such, the adverse influence of occluded pixels is suppressed in the cost fusion. The proposed framework Vis-MVSNet significantly improves depth accuracies in the scenes with severe occlusion. Extensive experiments are performed on DTU, BlendedMVS, and Tanks and Temples datasets to justify the effectiveness of the proposed framework.
Self-supervised learning has emerged as a powerful tool for depth and ego-motion estimation, leading to state-of-the-art results on benchmark datasets. However, one significant limitation shared by current methods is the assumption of a known parametric camera model -- usually the standard pinhole geometry -- leading to failure when applied to imaging systems that deviate significantly from this assumption (e.g., catadioptric cameras or underwater imaging). In this work, we show that self-supervision can be used to learn accurate depth and ego-motion estimation without prior knowledge of the camera model. Inspired by the geometric model of Grossberg and Nayar, we introduce Neural Ray Surfaces (NRS), convolutional networks that represent pixel-wise projection rays, approximating a wide range of cameras. NRS are fully differentiable and can be learned end-to-end from unlabeled raw videos. We demonstrate the use of NRS for self-supervised learning of visual odometry and depth estimation from raw videos obtained using a wide variety of camera systems, including pinhole, fisheye, and catadioptric.