Estimating depth from a single image represents an attractive alternative to more traditional approaches leveraging multiple cameras. In this field, deep learning yielded outstanding results at the cost of needing large amounts of data labeled with precise depth measurements for training. An issue softened by self-supervised approaches leveraging monocular sequences or stereo pairs in place of expensive ground truth depth annotations. This paper enables to further improve monocular depth estimation by integrating into existing self-supervised networks a geometrical prior. Specifically, we propose a sparsity-invariant autoencoder able to process the output of conventional visual odometry algorithms working in synergy with depth-from-mono networks. Experimental results on the KITTI dataset show that by exploiting the geometrical prior, our proposal: i) outperforms existing approaches in the literature and ii) couples well with both compact and complex depth-from-mono architectures, allowing for its deployment on high-end GPUs as well as on embedded devices (e.g., NVIDIA Jetson TX2).
In this paper, we address the problem of monocular depth estimation when only a limited number of training image-depth pairs are available. To achieve a high regression accuracy, the state-of-the-art estimation methods rely on CNNs trained with a large number of image-depth pairs, which are prohibitively costly or even infeasible to acquire. Aiming to break the curse of such expensive data collections, we propose a semi-supervised adversarial learning framework that only utilizes a small number of image-depth pairs in conjunction with a large number of easily-available monocular images to achieve high performance. In particular, we use one generator to regress the depth and two discriminators to evaluate the predicted depth , i.e., one inspects the image-depth pair while the other inspects the depth channel alone. These two discriminators provide their feedbacks to the generator as the loss to generate more realistic and accurate depth predictions. Experiments show that the proposed approach can (1) improve most state-of-the-art models on the NYUD v2 dataset by effectively leveraging additional unlabeled data sources; (2) reach state-of-the-art accuracy when the training set is small, e.g., on the Make3D dataset; (3) adapt well to an unseen new dataset (Make3D in our case) after training on an annotated dataset (KITTI in our case).
For ego-motion estimation, the feature representation of the scenes is crucial. Previous methods indicate that both the low-level and semantic feature-based methods can achieve promising results. Therefore, the incorporation of hierarchical feature representation may benefit from both methods. From this perspective, we propose a novel direct feature odometry framework, named DFO, for depth estimation and hierarchical feature representation learning from monocular videos. By exploiting the metric distance, our framework is able to learn the hierarchical feature representation without supervision. The pose is obtained with a coarse-to-fine approach from high-level to low-level features in enlarged feature maps. The pixel-level attention mask can be self-learned to provide the prior information. In contrast to the previous methods, our proposed method calculates the camera motion with a direct method rather than regressing the ego-motion from the pose network. With this approach, the consistency of the scale factor of translation can be constrained. Additionally, the proposed method is thus compatible with the traditional SLAM pipeline. Experiments on the KITTI dataset demonstrate the effectiveness of our method.
This paper addresses the problem of Monocular Depth Estimation (MDE). Existing approaches on MDE usually model it as a pixel-level regression problem, ignoring the underlying geometry property. We empirically find this may result in sub-optimal solution: while the predicted depth map presents small loss value in one specific view, it may exhibit large loss if viewed in different directions. In this paper, inspired by multi-view stereo (MVS), we propose an Adversarial View-Consistent Learning (AVCL) framework to force the estimated depth map to be all reasonable viewed from multiple views. To this end, we first design a differentiable depth map warping operation, which is end-to-end trainable, and then propose a pose generator to generate novel views for a given image in an adversarial manner. Collaborating with the differentiable depth map warping operation, the pose generator encourages the depth estimation network to learn from hard views, hence produce view-consistent depth maps . We evaluate our method on NYU Depth V2 dataset and the experimental results show promising performance gain upon state-of-the-art MDE approaches.
Color-coded aperture (CCA) methods can physically measure the depth of a scene given by physical cues from a single-shot image of a monocular camera. However, they are vulnerable to actual lens aberrations in real scenes because they assume an ideal lens for simplifying algorithms. In this paper, we propose physical cue-based deep learning for CCA photography. To address actual lens aberrations, we developed a deep deaberration network (DDN) that is additionally equipped with a self-attention mechanism of position and color channels to efficiently learn the lens aberration. Furthermore, a new Bayes L1 loss function based on Bayesian deep learning enables to handle the uncertainty of depth estimation more accurately. Quantitative and qualitative comparisons demonstrate that our method is superior to conventional methods including real outdoor scenes. Furthermore, compared to a long-baseline stereo camera, the proposed method provides an error-free depth map at close range, as there is no blind spot between the left and right cameras.
The 3D depth estimation and relative pose estimation problem within a decentralized architecture is a challenging problem that arises in missions that require coordination among multiple vision-controlled robots. The depth estimation problem aims at recovering the 3D information of the environment. The relative localization problem consists of estimating the relative pose between two robots, by sensing each other's pose or sharing information about the perceived environment. Most solutions for these problems use a set of discrete data without taking into account the chronological order of the events. This paper builds on recent results on continuous estimation to propose a framework that estimates the depth and relative pose between two non-holonomic vehicles. The basic idea consists in estimating the depth of the points by explicitly considering the dynamics of the camera mounted on a ground robot, and feeding the estimates of 3D points observed by both cameras in a filter that computes the relative pose between the robots. We evaluate the convergence for a set of simulated scenarios and show experimental results validating the proposed framework.
Running time of the light field depth estimation algorithms is typically high. This assessment is based on the computational complexity of existing methods and the large amounts of data involved. The aim of our work is to develop a simple and fast algorithm for accurate depth computation. In this context, we propose an approach, which involves Semi-Global Matching for the processing of light field images. It forms on comparison of pixels' correspondences with different metrics in the substantially bounded light field space. We show that our method is suitable for the fast production of a proper result in a variety of light field configurations
Monocular depth prediction plays a crucial role in understanding 3D scene geometry. Although recent methods have achieved impressive progress in evaluation metrics such as the pixel-wise relative error, most methods neglect the geometric constraints in the 3D space. In this work, we show the importance of the high-order 3D geometric constraints for depth prediction. By designing a loss term that enforces one simple type of geometric constraints, namely, virtual normal directions determined by randomly sampled three points in the reconstructed 3D space, we can considerably improve the depth prediction accuracy. Significantly, the byproduct of this predicted depth being sufficiently accurate is that we are now able to recover good 3D structures of the scene such as the point cloud and surface normal directly from the depth, eliminating the necessity of training new sub-models as was previously done. Experiments on two benchmarks: NYU Depth-V2 and KITTI demonstrate the effectiveness of our method and state-of-the-art performance.
We introduce MultiDepth, a novel training strategy and convolutional neural network (CNN) architecture that allows approaching single-image depth estimation (SIDE) as a multi-task problem. SIDE is an important part of road scene understanding. It, thus, plays a vital role in advanced driver assistance systems and autonomous vehicles. Best results for the SIDE task so far have been achieved using deep CNNs. However, optimization of regression problems, such as estimating depth, is still a challenging task. For the related tasks of image classification and semantic segmentation, numerous CNN-based methods with robust training behavior have been proposed. Hence, in order to overcome the notorious instability and slow convergence of depth value regression during training, MultiDepth makes use of depth interval classification as an auxiliary task. The auxiliary task can be disabled at test-time to predict continuous depth values using the main regression branch more efficiently. We applied MultiDepth to road scenes and present results on the KITTI depth prediction dataset. In experiments, we were able to show that end-to-end multi-task learning with both, regression and classification, is able to considerably improve training and yield more accurate results.
Depth cameras are utilized in many applications. Recently light field approaches are increasingly being used for depth computation. While these approaches demonstrate the technical feasibility, they can not be brought into real-world application, since they have both a high computation time as well as a large design. Exactly these two drawbacks are overcome in this paper. For the first time, we present a depth camera based on the light field principle, which provides real-time depth information as well as a compact design.
Autonomous vehicles and robots require a full scene understanding of the environment to interact with it. Such a perception typically incorporates pixel-wise knowledge of the depths and semantic labels for each image from a video sensor. Recent learning-based methods estimate both types of information independently using two separate CNNs. In this paper, we propose a model that is able to predict both outputs simultaneously, which leads to improved results and even reduced computational costs compared to independent estimation of depth and semantics. We also empirically prove that the CNN is capable of learning more meaningful and semantically richer features. Furthermore, our SDNet estimates the depth based on ordinal classification. On the basis of these two enhancements, our proposed method achieves state-of-the-art results in semantic segmentation and depth estimation from single monocular input images on two challenging datasets.
Estimating accurate depth from a single image is challenging because it is an ill-posed problem as infinitely many 3D scenes can be projected to the same 2D scene. However, recent works based on deep convolutional neural networks show great progress with plausible results. The convolutional neural networks are generally composed of two parts: an encoder for dense feature extraction and a decoder for predicting the desired depth. In the encoder-decoder schemes, repeated strided convolution and spatial pooling layers lower the spatial resolution of transitional outputs, and several techniques such as skip connections or multi-layer deconvolutional networks are adopted to recover the original resolution for effective dense prediction. In this paper, for more effective guidance of densely encoded features to the desired depth prediction, we propose a network architecture that utilizes novel local planar guidance layers located at multiple stages in the decoding phase. We show that the proposed method outperforms the state-of-the-art works with significant margin evaluating on challenging benchmarks. We also provide results from an ablation study to validate the effectiveness of the proposed method.
This work proposes a method for depth completion of sparse LiDAR data using a convolutional neural network which can be used to generate semi-dense depth maps and "almost" full 3D point-clouds with significantly lower root mean squared error (RMSE) over state-of-the-art methods. We add an "Error Prediction" unit to our network and present a novel and simple end-to-end method that learns to predict an error-map of depth regression task. An "almost" dense high-confidence/low-variance point-cloud is more valuable for safety-critical applications specifically real-world autonomous driving than a full point-cloud with high error rate and high error variance. Using our predicted error-map, we demonstrate that by up-filling a LiDAR point cloud from 18,000 points to 285,000 points, versus 300,000 points for full depth, we can reduce the RMSE error from 1004 to 399. This error is approximately 60% less than the state-of-the-art and 50% less than the state-of-the-art with RGB guidance (we did not use RGB guidance in our algorithm). In addition to analyzing our results on Kitti depth completion dataset, we also demonstrate the ability of our proposed method to extend to new tasks by deploying our "Error Prediction" unit to improve upon the state-of-the-art for monocular depth estimation. Codes and demo videos are available at http://github.com/hekmak/Conf-net.
We introduce Repetition-Reduction network (RRNet) for resource-constrained depth estimation, offering significantly improved efficiency in terms of computation, memory and energy consumption. The proposed method is based on repetition-reduction (RR) blocks. The RR blocks consist of the set of repeated convolutions and the residual connection layer that take place of the pointwise reduction layer with linear connection to the decoder. The RRNet help reduce memory usage and power consumption in the residual connections to the decoder layers. RRNet consumes approximately 3.84 times less energy and 3.06 times less meory and is approaximately 2.21 times faster, without increasing the demand on hardware resource relative to the baseline network (Godard et al, CVPR'17), outperforming current state-of-the-art lightweight architectures such as SqueezeNet, ShuffleNet, MobileNet and PyDNet.
In this technical report we investigate speed estimation of the ego-vehicle on the KITTI benchmark using state-of-the-art deep neural network based optical flow and single-view depth prediction methods. Using a straightforward intuitive approach and approximating a single scale factor, we evaluate several application schemes of the deep networks and formulate meaningful conclusions such as: combining depth information with optical flow improves speed estimation accuracy as opposed to using optical flow alone; the quality of the deep neural network methods influences speed estimation performance; using the depth and optical flow results from smaller crops of wide images degrades performance. With these observations in mind, we achieve a RMSE of less than 1 m/s for vehicle speed estimation using monocular images as input from recordings of the KITTI benchmark. Limitations and possible future directions are discussed as well.
Majority of state-of-the-art monocular depth estimation methods are supervised learning approaches. The success of such approaches heavily depends on the high-quality depth labels which are expensive to obtain. Some recent methods try to learn depth networks by leveraging unsupervised cues from monocular videos which are easier to acquire but less reliable. In this paper, we propose to resolve this dilemma by transferring knowledge from synthetic videos with easily obtainable ground-truth depth labels. Due to the stylish difference between synthetic and real images, we propose a temporally-consistent domain adaptation (TCDA) approach that simultaneously explores labels in the synthetic domain and temporal constraints in the videos to improve style transfer and depth prediction. Furthermore, we make use of the ground-truth optical flow and pose information in the synthetic data to learn moving mask and pose prediction networks. The learned moving masks can filter out moving regions that produces erroneous temporal constraints and the estimated poses provide better initializations for estimating temporal constraints. Experimental results demonstrate the effectiveness of our method and comparable performance against state-of-the-art.
Monocular depth estimation is an essential task for scene understanding. The underlying structure of objects and stuff in a complex scene is critical to recovering accurate and visually-pleasing depth maps. Global structure conveys scene layouts, while local structure reflects shape details. Recently developed approaches based on convolutional neural networks (CNNs) significantly improve the performance of depth estimation. However, few of them take into account multi-scale structures in complex scenes. In this paper, we propose a Structure-Aware Residual Pyramid Network (SARPN) to exploit multi-scale structures for accurate depth prediction. We propose a Residual Pyramid Decoder (RPD) which expresses global scene structure in upper levels to represent layouts, and local structure in lower levels to present shape details. At each level, we propose Residual Refinement Modules (RRM) that predict residual maps to progressively add finer structures on the coarser structure predicted at the upper level. In order to fully exploit multi-scale image features, an Adaptive Dense Feature Fusion (ADFF) module, which adaptively fuses effective features from all scales for inferring structures of each scale, is introduced. Experiment results on the challenging NYU-Depth v2 dataset demonstrate that our proposed approach achieves state-of-the-art performance in both qualitative and quantitative evaluation. The code is available at https://github.com/Xt-Chen/SARPN.
The success of monocular depth estimation relies on large and diverse training sets. Due to the challenges associated with acquiring dense ground-truth depth across different environments at scale, a number of datasets with distinct characteristics and biases have emerged. We develop tools that enable mixing multiple datasets during training, even if their annotations are incompatible. In particular, we propose a robust training objective that is invariant to changes in depth range and scale, advocate the use of principled multi-objective learning to combine data from different sources, and highlight the importance of pretraining encoders on auxiliary tasks. Armed with these tools, we experiment with five diverse training datasets, including a new, massive data source: 3D films. To demonstrate the generalization power of our approach we use zero-shot cross-dataset transfer}, i.e. we evaluate on datasets that were not seen during training. The experiments confirm that mixing data from complementary sources greatly improves monocular depth estimation. Our approach clearly outperforms competing methods across diverse datasets, setting a new state of the art for monocular depth estimation. Some results are shown in the supplementary video at https://youtu.be/D46FzVyL9I8
In this work we present a method to train a plane-aware convolutional neural network for dense depth and surface normal estimation as well as plane boundaries from a single indoor $360^\circ$ image. Using our proposed loss function, our network outperforms existing methods for single-view, indoor, omnidirectional depth estimation and provides an initial benchmark for surface normal prediction from $360^\circ$ images. Our improvements are due to the use of a novel plane-aware loss that leverages principal curvature as an indicator of planar boundaries. We also show that including geodesic coordinate maps as network priors provides a significant boost in surface normal prediction accuracy. Finally, we demonstrate how we can combine our network's outputs to generate high quality 3D "pop-up" models of indoor scenes.
Medical endoscopy remains a challenging application for simultaneous localization and mapping (SLAM) due to the sparsity of image features and size constraints that prevent direct depth-sensing. We present a SLAM approach that incorporates depth predictions made by an adversarially-trained convolutional neural network (CNN) applied to monocular endoscopy images. The depth network is trained with synthetic images of a simple colon model, and then fine-tuned with domain-randomized, photorealistic images rendered from computed tomography measurements of human colons. Each image is paired with an error-free depth map for supervised adversarial learning. Monocular RGB images are then fused with corresponding depth predictions, enabling dense reconstruction and mosaicing as an endoscope is advanced through the gastrointestinal tract. Our preliminary results demonstrate that incorporating monocular depth estimation into a SLAM architecture can enable dense reconstruction of endoscopic scenes.