Multitask learning is a common approach in machine learning, which allows to train multiple objectives with a shared architecture. It has been shown that by training multiple tasks together inference time and compute resources can be saved, while the objectives performance remains on a similar or even higher level. However, in perception related multitask networks only closely related tasks can be found, such as object detection, instance and semantic segmentation or depth estimation. Multitask networks with diverse tasks and their effects with respect to efficiency on one another are not well studied. In this paper we augment the CenterNet anchor-free approach for training multiple diverse perception related tasks together, including the task of object detection and semantic segmentation as well as human pose estimation. We refer to this DNN as Multitask-CenterNet (MCN). Additionally, we study different MCN settings for efficiency. The MCN can perform several tasks at once while maintaining, and in some cases even exceeding, the performance values of its corresponding single task networks. More importantly, the MCN architecture decreases inference time and reduces network size when compared to a composition of single task networks.
While self-supervised monocular depth estimation in driving scenarios has achieved comparable performance to supervised approaches, violations of the static world assumption can still lead to erroneous depth predictions of traffic participants, posing a potential safety issue. In this paper, we present R4Dyn, a novel set of techniques to use cost-efficient radar data on top of a self-supervised depth estimation framework. In particular, we show how radar can be used during training as weak supervision signal, as well as an extra input to enhance the estimation robustness at inference time. Since automotive radars are readily available, this allows to collect training data from a variety of existing vehicles. Moreover, by filtering and expanding the signal to make it compatible with learning-based approaches, we address radar inherent issues, such as noise and sparsity. With R4Dyn we are able to overcome a major limitation of self-supervised depth estimation, i.e. the prediction of traffic participants. We substantially improve the estimation on dynamic objects, such as cars by 37% on the challenging nuScenes dataset, hence demonstrating that radar is a valuable additional sensor for monocular depth estimation in autonomous vehicles.
Scene understanding is crucial for autonomous systems which intend to operate in the real world. Single task vision networks extract information only based on some aspects of the scene. In multi-task learning (MTL), on the other hand, these single tasks are jointly learned, thereby providing an opportunity for tasks to share information and obtain a more comprehensive understanding. To this end, we develop UniNet, a unified scene understanding network that accurately and efficiently infers vital vision tasks including object detection, semantic segmentation, instance segmentation, monocular depth estimation, and monocular instance depth prediction. As these tasks look at different semantic and geometric information, they can either complement or conflict with each other. Therefore, understanding inter-task relationships can provide useful cues to enable complementary information sharing. We evaluate the task relationships in UniNet through the lens of adversarial attacks based on the notion that they can exploit learned biases and task interactions in the neural network. Extensive experiments on the Cityscapes dataset, using untargeted and targeted attacks reveal that semantic tasks strongly interact amongst themselves, and the same holds for geometric tasks. Additionally, we show that the relationship between semantic and geometric tasks is asymmetric and their interaction becomes weaker as we move towards higher-level representations.
Monocular depth estimation aims at predicting depth from a single image or video. Recently, self-supervised methods draw much attention since they are free of depth annotations and achieve impressive performance on several daytime benchmarks. However, they produce weird outputs in more challenging nighttime scenarios because of low visibility and varying illuminations, which bring weak textures and break brightness-consistency assumption, respectively. To address these problems, in this paper we propose a novel framework with several improvements: (1) we introduce Priors-Based Regularization to learn distribution knowledge from unpaired depth maps and prevent model from being incorrectly trained; (2) we leverage Mapping-Consistent Image Enhancement module to enhance image visibility and contrast while maintaining brightness consistency; and (3) we present Statistics-Based Mask strategy to tune the number of removed pixels within textureless regions, using dynamic statistics. Experimental results demonstrate the effectiveness of each component. Meanwhile, our framework achieves remarkable improvements and state-of-the-art results on two nighttime datasets.
Bird's-Eye-View (BEV) maps have emerged as one of the most powerful representations for scene understanding due to their ability to provide rich spatial context while being easy to interpret and process. Such maps have found use in many real-world tasks that extensively rely on accurate scene segmentation as well as object instance identification in the BEV space for their operation. However, existing segmentation algorithms only predict the semantics in the BEV space, which limits their use in applications where the notion of object instances is also critical. In this work, we present the first BEV panoptic segmentation approach for directly predicting dense panoptic segmentation maps in the BEV, given a single monocular image in the frontal view (FV). Our architecture follows the top-down paradigm and incorporates a novel dense transformer module consisting of two distinct transformers that learn to independently map vertical and flat regions in the input image from the FV to the BEV. Additionally, we derive a mathematical formulation for the sensitivity of the FV-BEV transformation which allows us to intelligently weight pixels in the BEV space to account for the varying descriptiveness across the FV image. Extensive evaluations on the KITTI-360 and nuScenes datasets demonstrate that our approach exceeds the state-of-the-art in the PQ metric by 3.61 pp and 4.93 pp respectively.
Real-world perception systems in many cases build on hardware with limited resources to adhere to cost and power limitations of their carrying system. Deploying deep neural networks on resource-constrained hardware became possible with model compression techniques, as well as efficient and hardware-aware architecture design. However, model adaptation is additionally required due to the diverse operation environments. In this work, we address the problem of training deep neural networks on resource-constrained hardware in the context of visual domain adaptation. We select the task of monocular depth estimation where our goal is to transform a pre-trained model to the target's domain data. While the source domain includes labels, we assume an unlabelled target domain, as it happens in real-world applications. Then, we present an adversarial learning approach that is adapted for training on the device with limited resources. Since visual domain adaptation, i.e. neural network training, has not been previously explored for resource-constrained hardware, we present the first feasibility study for image-based depth estimation. Our experiments show that visual domain adaptation is relevant only for efficient network architectures and training sets at the order of a few hundred samples. Models and code are publicly available.
We design a multiscopic vision system that utilizes a low-cost monocular RGB camera to acquire accurate depth estimation. Unlike multi-view stereo with images captured at unconstrained camera poses, the proposed system controls the motion of a camera to capture a sequence of images in horizontally or vertically aligned positions with the same parallax. In this system, we propose a new heuristic method and a robust learning-based method to fuse multiple cost volumes between the reference image and its surrounding images. To obtain training data, we build a synthetic dataset with multiscopic images. The experiments on the real-world Middlebury dataset and real robot demonstration show that our multiscopic vision system outperforms traditional two-frame stereo matching methods in depth estimation. Our code and dataset are available at https://sites.google.com/view/multiscopic.
Dynamic objects in the environment, such as people and other agents, lead to challenges for existing simultaneous localization and mapping (SLAM) approaches. To deal with dynamic environments, computer vision researchers usually apply some learning-based object detectors to remove these dynamic objects. However, these object detectors are computationally too expensive for mobile robot on-board processing. In practical applications, these objects output noisy sounds that can be effectively detected by on-board sound source localization. The directional information of the sound source object can be efficiently obtained by direction of sound arrival (DoA) estimation, but depth estimation is difficult. Therefore, in this paper, we propose a novel audio-visual fusion approach that fuses sound source direction into the RGB-D image and thus removes the effect of dynamic obstacles on the multi-robot SLAM system. Experimental results of multi-robot SLAM in different dynamic environments show that the proposed method uses very small computational resources to obtain very stable self-localization results.
We present a method to estimate depth of a dynamic scene, containing arbitrary moving objects, from an ordinary video captured with a moving camera. We seek a geometrically and temporally consistent solution to this underconstrained problem: the depth predictions of corresponding points across frames should induce plausible, smooth motion in 3D. We formulate this objective in a new test-time training framework where a depth-prediction CNN is trained in tandem with an auxiliary scene-flow prediction MLP over the entire input video. By recursively unrolling the scene-flow prediction MLP over varying time steps, we compute both short-range scene flow to impose local smooth motion priors directly in 3D, and long-range scene flow to impose multi-view consistency constraints with wide baselines. We demonstrate accurate and temporally coherent results on a variety of challenging videos containing diverse moving objects (pets, people, cars), as well as camera motion. Our depth maps give rise to a number of depth-and-motion aware video editing effects such as object and lighting insertion.
Good quality reconstruction and comprehension of a scene rely on 3D estimation methods. The 3D information was usually obtained from images by stereo-photogrammetry, but deep learning has recently provided us with excellent results for monocular depth estimation. Building up a sufficiently large and rich training dataset to achieve these results requires onerous processing. In this paper, we address the problem of learning outdoor 3D point cloud from monocular data using a sparse ground-truth dataset. We propose Pix2Point, a deep learning-based approach for monocular 3D point cloud prediction, able to deal with complete and challenging outdoor scenes. Our method relies on a 2D-3D hybrid neural network architecture, and a supervised end-to-end minimisation of an optimal transport divergence between point clouds. We show that, when trained on sparse point clouds, our simple promising approach achieves a better coverage of 3D outdoor scenes than efficient monocular depth methods.
As a crucial task of autonomous driving, 3D object detection has made great progress in recent years. However, monocular 3D object detection remains a challenging problem due to the unsatisfactory performance in depth estimation. Most existing monocular methods typically directly regress the scene depth while ignoring important relationships between the depth and various geometric elements (e.g. bounding box sizes, 3D object dimensions, and object poses). In this paper, we propose to learn geometry-guided depth estimation with projective modeling to advance monocular 3D object detection. Specifically, a principled geometry formula with projective modeling of 2D and 3D depth predictions in the monocular 3D object detection network is devised. We further implement and embed the proposed formula to enable geometry-aware deep representation learning, allowing effective 2D and 3D interactions for boosting the depth estimation. Moreover, we provide a strong baseline through addressing substantial misalignment between 2D annotation and projected boxes to ensure robust learning with the proposed geometric formula. Experiments on the KITTI dataset show that our method remarkably improves the detection performance of the state-of-the-art monocular-based method without extra data by 2.80% on the moderate test setting. The model and code will be released at https://github.com/YinminZhang/MonoGeo.
Depth completion deals with the problem of recovering dense depth maps from sparse ones, where color images are often used to facilitate this task. Recent approaches mainly focus on image guided learning frameworks to predict dense depth. However, blurry guidance in the image and unclear structure in the depth still impede the performance of the image guided frameworks. To tackle these problems, we explore a repetitive design in our image guided network to gradually and sufficiently recover depth values. Specifically, the repetition is embodied in both the image guidance branch and depth generation branch. In the former branch, we design a repetitive hourglass network to extract discriminative image features of complex environments, which can provide powerful contextual instruction for depth prediction. In the latter branch, we introduce a repetitive guidance module based on dynamic convolution, in which an efficient convolution factorization is proposed to simultaneously reduce its complexity and progressively model high-frequency structures. Extensive experiments show that our method achieves superior or competitive results on KITTI benchmark and NYUv2 dataset.
Monocular depth estimation and semantic segmentation are two fundamental goals of scene understanding. Due to the advantages of task interaction, many works study the joint task learning algorithm. However, most existing methods fail to fully leverage the semantic labels, ignoring the provided context structures and only using them to supervise the prediction of segmentation split, which limit the performance of both tasks. In this paper, we propose a network injected with contextual information (CI-Net) to solve the problem. Specifically, we introduce self-attention block in the encoder to generate attention map. With supervision from the ideal attention map created by semantic label, the network is embedded with contextual information so that it could understand scene better and utilize correlated features to make accurate prediction. Besides, a feature sharing module is constructed to make the task-specific features deeply fused and a consistency loss is devised to make the features mutually guided. We evaluate the proposed CI-Net on the NYU-Depth-v2 and SUN-RGBD datasets. The experimental results validate that our proposed CI-Net could effectively improve the accuracy of semantic segmentation and depth estimation.
Geometry Projection is a powerful depth estimation method in monocular 3D object detection. It estimates depth dependent on heights, which introduces mathematical priors into the deep model. But projection process also introduces the error amplification problem, in which the error of the estimated height will be amplified and reflected greatly at the output depth. This property leads to uncontrollable depth inferences and also damages the training efficiency. In this paper, we propose a Geometry Uncertainty Projection Network (GUP Net) to tackle the error amplification problem at both inference and training stages. Specifically, a GUP module is proposed to obtains the geometry-guided uncertainty of the inferred depth, which not only provides high reliable confidence for each depth but also benefits depth learning. Furthermore, at the training stage, we propose a Hierarchical Task Learning strategy to reduce the instability caused by error amplification. This learning algorithm monitors the learning situation of each task by a proposed indicator and adaptively assigns the proper loss weights for different tasks according to their pre-tasks situation. Based on that, each task starts learning only when its pre-tasks are learned well, which can significantly improve the stability and efficiency of the training process. Extensive experiments demonstrate the effectiveness of the proposed method. The overall model can infer more reliable object depth than existing methods and outperforms the state-of-the-art image-based monocular 3D detectors by 3.74% and 4.7% AP40 of the car and pedestrian categories on the KITTI benchmark.
3D object detection is an important capability needed in various practical applications such as driver assistance systems. Monocular 3D detection, as a representative general setting among image-based approaches, provides a more economical solution than conventional settings relying on LiDARs but still yields unsatisfactory results. This paper first presents a systematic study on this problem. We observe that the current monocular 3D detection can be simplified as an instance depth estimation problem: The inaccurate instance depth blocks all the other 3D attribute predictions from improving the overall detection performance. Moreover, recent methods directly estimate the depth based on isolated instances or pixels while ignoring the geometric relations across different objects. To this end, we construct geometric relation graphs across predicted objects and use the graph to facilitate depth estimation. As the preliminary depth estimation of each instance is usually inaccurate in this ill-posed setting, we incorporate a probabilistic representation to capture the uncertainty. It provides an important indicator to identify confident predictions and further guide the depth propagation. Despite the simplicity of the basic idea, our method, PGD, obtains significant improvements on KITTI and nuScenes benchmarks, achieving 1st place out of all monocular vision-only methods while still maintaining real-time efficiency. Code and models will be released at https://github.com/open-mmlab/mmdetection3d.
Road detection is a critically important task for self-driving cars. By employing LiDAR data, recent works have significantly improved the accuracy of road detection. Relying on LiDAR sensors limits the wide application of those methods when only cameras are available. In this paper, we propose a novel road detection approach with RGB being the only input during inference. Specifically, we exploit pseudo-LiDAR using depth estimation, and propose a feature fusion network where RGB and learned depth information are fused for improved road detection. To further optimize the network structure and improve the efficiency of the network. we search for the network structure of the feature fusion module using NAS techniques. Finally, be aware of that generating pseudo-LiDAR from RGB via depth estimation introduces extra computational costs and relies on depth estimation networks, we design a modality distillation strategy and leverage it to further free our network from these extra computational cost and dependencies during inference. The proposed method achieves state-of-the-art performance on two challenging benchmarks, KITTI and R2D.
Current geometry-based monocular 3D object detection models can efficiently detect objects by leveraging perspective geometry, but their performance is limited due to the absence of accurate depth information. Though this issue can be alleviated in a depth-based model where a depth estimation module is plugged to predict depth information before 3D box reasoning, the introduction of such module dramatically reduces the detection speed. Instead of training a costly depth estimator, we propose a rendering module to augment the training data by synthesizing images with virtual-depths. The rendering module takes as input the RGB image and its corresponding sparse depth image, outputs a variety of photo-realistic synthetic images, from which the detection model can learn more discriminative features to adapt to the depth changes of the objects. Besides, we introduce an auxiliary module to improve the detection model by jointly optimizing it through a depth estimation task. Both modules are working in the training time and no extra computation will be introduced to the detection model. Experiments show that by working with our proposed modules, a geometry-based model can represent the leading accuracy on the KITTI 3D detection benchmark.
With the development of computational intelligence algorithms, unsupervised monocular depth and pose estimation framework, which is driven by warped photometric consistency, has shown great performance in the daytime scenario. While in some challenging environments, like night and rainy night, the essential photometric consistency hypothesis is untenable because of the complex lighting and reflection, so that the above unsupervised framework cannot be directly applied to these complex scenarios. In this paper, we investigate the problem of unsupervised monocular depth estimation in highly complex scenarios and address this challenging problem by adopting an image transfer-based domain adaptation framework. We adapt the depth model trained on day-time scenarios to be applicable to night-time scenarios, and constraints on both feature space and output space promote the framework to learn the key features for depth decoding. Meanwhile, we further tackle the effects of unstable image transfer quality on domain adaptation, and an image adaptation approach is proposed to evaluate the quality of transferred images and re-weight the corresponding losses, so as to improve the performance of the adapted depth model. Extensive experiments show the effectiveness of the proposed unsupervised framework in estimating the dense depth map from highly complex images.
Depth map super-resolution is a task with high practical application requirements in the industry. Existing color-guided depth map super-resolution methods usually necessitate an extra branch to extract high-frequency detail information from RGB image to guide the low-resolution depth map reconstruction. However, because there are still some differences between the two modalities, direct information transmission in the feature dimension or edge map dimension cannot achieve satisfactory result, and may even trigger texture copying in areas where the structures of the RGB-D pair are inconsistent. Inspired by the multi-task learning, we propose a joint learning network of depth map super-resolution (DSR) and monocular depth estimation (MDE) without introducing additional supervision labels. For the interaction of two subnetworks, we adopt a differentiated guidance strategy and design two bridges correspondingly. One is the high-frequency attention bridge (HABdg) designed for the feature encoding process, which learns the high-frequency information of the MDE task to guide the DSR task. The other is the content guidance bridge (CGBdg) designed for the depth map reconstruction process, which provides the content guidance learned from DSR task for MDE task. The entire network architecture is highly portable and can provide a paradigm for associating the DSR and MDE tasks. Extensive experiments on benchmark datasets demonstrate that our method achieves competitive performance. Our code and models are available at https://rmcong.github.io/proj_BridgeNet.html.
Self-supervised depth estimation for indoor environments is more challenging than its outdoor counterpart in at least the following two aspects: (i) the depth range of indoor sequences varies a lot across different frames, making it difficult for the depth network to induce consistent depth cues, whereas the maximum distance in outdoor scenes mostly stays the same as the camera usually sees the sky; (ii) the indoor sequences contain much more rotational motions, which cause difficulties for the pose network, while the motions of outdoor sequences are pre-dominantly translational, especially for driving datasets such as KITTI. In this paper, special considerations are given to those challenges and a set of good practices are consolidated for improving the performance of self-supervised monocular depth estimation in indoor environments. The proposed method mainly consists of two novel modules, \ie, a depth factorization module and a residual pose estimation module, each of which is designed to respectively tackle the aforementioned challenges. The effectiveness of each module is shown through a carefully conducted ablation study and the demonstration of the state-of-the-art performance on three indoor datasets, \ie, EuRoC, NYUv2, and 7-scenes.