In this paper, we introduce a new dataset, named InstaOrder, that can be used to understand the geometrical relationships of instances in an image. The dataset consists of 2.9M annotations of geometric orderings for class-labeled instances in 101K natural scenes. The scenes were annotated by 3,659 crowd-workers regarding (1) occlusion order that identifies occluder/occludee and (2) depth order that describes ordinal relations that consider relative distance from the camera. The dataset provides joint annotation of two kinds of orderings for the same instances, and we discover that the occlusion order and depth order are complementary. We also introduce a geometric order prediction network called InstaOrderNet, which is superior to state-of-the-art approaches. Moreover, we propose a dense depth prediction network called InstaDepthNet that uses auxiliary geometric order loss to boost the accuracy of the state-of-the-art depth prediction approach, MiDaS [56].
Recent years have seen impressive progress in visual recognition on many benchmarks, however, generalization to the real-world in out-of-distribution setting remains a significant challenge. A state-of-the-art method for robust visual recognition is model ensembling. however, recently it was shown that similarly competitive results could be achieved with a much smaller cost, by using multi-input multi-output architecture (MIMO). In this work, a generalization of the MIMO approach is applied to the task of object detection using the general-purpose Faster R-CNN model. It was shown that using the MIMO framework allows building strong feature representation and obtains very competitive accuracy when using just two input/output pairs. Furthermore, it adds just 0.5\% additional model parameters and increases the inference time by 15.9\% when compared to the standard Faster R-CNN. It also works comparably to, or outperforms the Deep Ensemble approach in terms of model accuracy, robustness to out-of-distribution setting, and uncertainty calibration when the same number of predictions is used. This work opens up avenues for applying the MIMO approach in other high-level tasks such as semantic segmentation and depth estimation.
Real-scale scene flow estimation has become increasingly important for 3D computer vision. Some works successfully estimate real-scale 3D scene flow with LiDAR. However, these ubiquitous and expensive sensors are still unlikely to be equipped widely for real application. Other works use monocular images to estimate scene flow, but their scene flow estimations are normalized with scale ambiguity, where additional depth or point cloud ground truth are required to recover the real scale. Even though they perform well in 2D, these works do not provide accurate and reliable 3D estimates. We present a deep learning architecture on permutohedral lattice - MonoPLFlowNet. Different from all previous works, our MonoPLFlowNet is the first work where only two consecutive monocular images are used as input, while both depth and 3D scene flow are estimated in real scale. Our real-scale scene flow estimation outperforms all state-of-the-art monocular-image based works recovered to real scale by ground truth, and is comparable to LiDAR approaches. As a by-product, our real-scale depth estimation also outperforms other state-of-the-art works.
Mapping and 3D detection are two major issues in vision-based robotics, and self-driving. While previous works only focus on each task separately, we present an innovative and efficient multi-task deep learning framework (SM3D) for Simultaneous Mapping and 3D Detection by bridging the gap with robust depth estimation and "Pseudo-LiDAR" point cloud for the first time. The Mapping module takes consecutive monocular frames to generate depth and pose estimation. In 3D Detection module, the depth estimation is projected into 3D space to generate "Pseudo-LiDAR" point cloud, where LiDAR-based 3D detector can be leveraged on point cloud for vehicular 3D detection and localization. By end-to-end training of both modules, the proposed mapping and 3D detection method outperforms the state-of-the-art baseline by 10.0% and 13.2% in accuracy, respectively. While achieving better accuracy, our monocular multi-task SM3D is more than 2 times faster than pure stereo 3D detector, and 18.3% faster than using two modules separately.
Stereo vision is an effective technique for depth estimation with broad applicability in autonomous urban and highway driving. While various deep learning-based approaches have been developed for stereo, the input data from a binocular setup with a fixed baseline are limited. Addressing such a problem, we present an end-to-end network for processing the data from a trinocular setup, which is a combination of a narrow and a wide stereo pair. In this design, two pairs of binocular data with a common reference image are treated with shared weights of the network and a mid-level fusion. We also propose a Guided Addition method for merging the 4D data of the two baselines. Additionally, an iterative sequential self-supervised and supervised learning on real and synthetic datasets is presented, making the training of the trinocular system practical with no need to ground-truth data of the real dataset. Experimental results demonstrate that the trinocular disparity network surpasses the scenario where individual pairs are fed into a similar architecture. Code and dataset: https://github.com/cogsys-tuebingen/tristereonet.
Recent visual odometry (VO) methods incorporating geometric algorithm into deep-learning architecture have shown outstanding performance on the challenging monocular VO task. Despite encouraging results are shown, previous methods ignore the requirement of generalization capability under noisy environment and various scenes. To address this challenging issue, this work first proposes a novel optical flow network (PANet). Compared with previous methods that predict optical flow as a direct regression task, our PANet computes optical flow by predicting it into the discrete position space with optical flow probability volume, and then converting it to optical flow. Next, we improve the bundle adjustment module to fit the self-supervised training pipeline by introducing multiple sampling, ego-motion initialization, dynamic damping factor adjustment, and Jacobi matrix weighting. In addition, a novel normalized photometric loss function is advanced to improve the depth estimation accuracy. The experiments show that the proposed system not only achieves comparable performance with other state-of-the-art self-supervised learning-based methods on the KITTI dataset, but also significantly improves the generalization capability compared with geometry-based, learning-based and hybrid VO systems on the noisy KITTI and the challenging outdoor (KAIST) scenes.
One of the key elements of reconstructing a 3D mesh from a monocular video is generating every frame's depth map. However, in the application of colonoscopy video reconstruction, producing good-quality depth estimation is challenging. Neural networks can be easily fooled by photometric distractions or fail to capture the complex shape of the colon surface, predicting defective shapes that result in broken meshes. Aiming to fundamentally improve the depth estimation quality for colonoscopy 3D reconstruction, in this work we have designed a set of training losses to deal with the special challenges of colonoscopy data. For better training, a set of geometric consistency objectives was developed, using both depth and surface normal information. Also, the classic photometric loss was extended with feature matching to compensate for illumination noise. With the training losses powerful enough, our self-supervised framework named ColDE is able to produce better depth maps of colonoscopy data as compared to the previous work utilizing prior depth knowledge. Used in reconstruction, our network is able to reconstruct good-quality colon meshes in real-time without any post-processing, making it the first to be clinically applicable.
We present a method for learning a generative 3D model based on neural radiance fields, trained solely from data with only single views of each object. While generating realistic images is no longer a difficult task, producing the corresponding 3D structure such that they can be rendered from different views is non-trivial. We show that, unlike existing methods, one does not need multi-view data to achieve this goal. Specifically, we show that by reconstructing many images aligned to an approximate canonical pose with a single network conditioned on a shared latent space, you can learn a space of radiance fields that models shape and appearance for a class of objects. We demonstrate this by training models to reconstruct object categories using datasets that contain only one view of each subject without depth or geometry information. Our experiments show that we achieve state-of-the-art results in novel view synthesis and high-quality results for monocular depth prediction.
Temporally consistent depth estimation is crucial for online applications such as augmented reality. While stereo depth estimation has received substantial attention as a promising way to generate 3D information, there is relatively little work focused on maintaining temporal stability. Indeed, based on our analysis, current techniques still suffer from poor temporal consistency. Stabilizing depth temporally in dynamic scenes is challenging due to concurrent object and camera motion. In an online setting, this process is further aggravated because only past frames are available. We present a framework named Consistent Online Dynamic Depth (CODD) to produce temporally consistent depth estimates in dynamic scenes in an online setting. CODD augments per-frame stereo networks with novel motion and fusion networks. The motion network accounts for dynamics by predicting a per-pixel SE3 transformation and aligning the observations. The fusion network improves temporal depth consistency by aggregating the current and past estimates. We conduct extensive experiments and demonstrate quantitatively and qualitatively that CODD outperforms competing methods in terms of temporal consistency and performs on par in terms of per-frame accuracy.
We propose SUB-Depth, a universal multi-task training framework for self-supervised monocular depth estimation (SDE). Depth models trained with SUB-Depth outperform the same models trained in a standard single-task SDE framework. By introducing an additional self-distillation task into a standard SDE training framework, SUB-Depth trains a depth network, not only to predict the depth map for an image reconstruction task, but also to distill knowledge from a trained teacher network with unlabelled data. To take advantage of this multi-task setting, we propose homoscedastic uncertainty formulations for each task to penalize areas likely to be affected by teacher network noise, or violate SDE assumptions. We present extensive evaluations on KITTI to demonstrate the improvements achieved by training a range of existing networks using the proposed framework, and we achieve state-of-the-art performance on this task. Additionally, SUB-Depth enables models to estimate uncertainty on depth output.
Depth estimation attracts widespread attention in the computer vision community. However, it is still quite difficult to recover an accurate depth map using only one RGB image. We observe a phenomenon that existing methods tend to fail in different cases, caused by differences in network architecture, loss function and so on. In this work, we investigate into the phenomenon and propose to integrate the strengths of multiple weak depth predictor to build a comprehensive and accurate depth predictor, which is critical for many real-world applications, e.g., 3D reconstruction. Specifically, we construct multiple base (weak) depth predictors by utilizing different Transformer-based and convolutional neural network (CNN)-based architectures. Transformer establishes long-range correlation while CNN preserves local information ignored by Transformer due to the spatial inductive bias. Therefore, the coupling of Transformer and CNN contributes to the generation of complementary depth estimates, which are essential to achieve a comprehensive depth predictor. Then, we design mixers to learn from multiple weak predictions and adaptively fuse them into a strong depth estimate. The resultant model, which we refer to as Transformer-assisted depth ensembles (TEDepth). On the standard NYU-Depth-v2 and KITTI datasets, we thoroughly explore how the neural ensembles affect the depth estimation and demonstrate that our TEDepth achieves better results than previous state-of-the-art approaches. To validate the generalizability across cameras, we directly apply the models trained on NYU-Depth-v2 to the SUN RGB-D dataset without any fine-tuning, and the superior results emphasize its strong generalizability.
Photometric consistency loss is one of the representative objective functions commonly used for self-supervised monocular depth estimation. However, this loss often causes unstable depth predictions in textureless or occluded regions due to incorrect guidance. Recent self-supervised learning approaches tackle this issue by utilizing feature representations explicitly learned from auto-encoders, expecting better discriminability than the input image. Despite the use of auto-encoded features, we observe that the method does not embed features as discriminative as auto-encoded features. In this paper, we propose residual guidance loss that enables the depth estimation network to embed the discriminative feature by transferring the discriminability of auto-encoded features. We conducted experiments on the KITTI benchmark and verified our method's superiority and orthogonality on other state-of-the-art methods.
The ability of accurate depth prediction by a convolutional neural network (CNN) is a major challenge for its wide use in practical visual simultaneous localization and mapping (SLAM) applications, such as enhanced camera tracking and dense mapping. This paper is set out to answer the following question: Can we tune a depth prediction CNN with the help of a visual SLAM algorithm even if the CNN is not trained for the current operating environment in order to benefit the SLAM performance? To this end, we propose a novel online adaptation framework consisting of two complementary processes: a SLAM algorithm that is used to generate keyframes to fine-tune the depth prediction and another algorithm that uses the online adapted depth to improve map quality. Once the potential noisy map points are removed, we perform global photometric bundle adjustment (BA) to improve the overall SLAM performance. Experimental results on both benchmark datasets and a real robot in our own experimental environments show that our proposed method improves the overall SLAM accuracy. While regularization has been shown to be effective in multi-task classification problems, we present experimental results and an ablation study to show the effectiveness of regularization in preventing catastrophic forgetting in the online adaptation of depth prediction, a single-task regression problem. In addition, we compare our online adaptation framework against the state-of-the-art pre-trained depth prediction CNNs to show that our online adapted depth prediction CNN outperforms the depth prediction CNNs that have been trained on a large collection of datasets.
Performing single image holistic understanding and 3D reconstruction is a central task in computer vision. This paper presents an integrated system that performs dense scene labeling, object detection, instance segmentation, depth estimation, 3D shape reconstruction, and 3D layout estimation for indoor and outdoor scenes from a single RGB image. We name our system panoptic 3D parsing (Panoptic3D) in which panoptic segmentation ("stuff" segmentation and "things" detection/segmentation) with 3D reconstruction is performed. We design a stage-wise system, Panoptic3D (stage-wise), where a complete set of annotations is absent. Additionally, we present an end-to-end pipeline, Panoptic3D (end-to-end), trained on a synthetic dataset with a full set of annotations. We show results on both indoor (3D-FRONT) and outdoor (COCO and Cityscapes) scenes. Our proposed panoptic 3D parsing framework points to a promising direction in computer vision. Panoptic3D can be applied to a variety of applications, including autonomous driving, mapping, robotics, design, computer graphics, robotics, human-computer interaction, and augmented reality.
Originally developed in fields such as robotics and autonomous driving with image-based navigation in mind, deep learning-based single-image depth estimation (SIDE) has found great interest in the wider image analysis community. Remote sensing is no exception, as the possibility to estimate height maps from single aerial or satellite imagery bears great potential in the context of topographic reconstruction. A few pioneering investigations have demonstrated the general feasibility of single image height prediction from optical remote sensing images and motivate further studies in that direction. With this paper, we present the first-ever demonstration of deep learning-based single image height prediction for the other important sensor modality in remote sensing: synthetic aperture radar (SAR) data. Besides the adaptation of a convolutional neural network (CNN) architecture for SAR intensity images, we present a workflow for the generation of training data, and extensive experimental results for different SAR imaging modes and test sites. Since we put a particular emphasis on transferability, we are able to confirm that deep learning-based single-image height estimation is not only possible, but also transfers quite well to unseen data, even if acquired by different imaging modes and imaging parameters.
Determining the distance between the objects in a scene and the camera sensor from 2D images is feasible by estimating depth images using stereo cameras or 3D cameras. The outcome of depth estimation is relative distances that can be used to calculate absolute distances to be applicable in reality. However, distance estimation is very challenging using 2D monocular cameras. This paper presents a deep learning framework that consists of two deep networks for depth estimation and object detection using a single image. Firstly, objects in the scene are detected and localized using the You Only Look Once (YOLOv5) network. In parallel, the estimated depth image is computed using a deep autoencoder network to detect the relative distances. The proposed object detection based YOLO was trained using a supervised learning technique, in turn, the network of depth estimation was self-supervised training. The presented distance estimation framework was evaluated on real images of outdoor scenes. The achieved results show that the proposed framework is promising and it yields an accuracy of 96% with RMSE of 0.203 of the correct absolute distance.
Perceiving 3D information is of paramount importance in many applications of computer vision. Recent advances in monocular depth estimation have shown that gaining such knowledge from a single camera input is possible by training deep neural networks to predict inverse depth and pose, without the necessity of ground truth data. The majority of such approaches, however, require camera parameters to be fed explicitly during training. As a result, image sequences from wild cannot be used during training. While there exist methods which also predict camera intrinsics, their performance is not on par with novel methods taking camera parameters as input. In this work, we propose a method for implicit estimation of pinhole camera intrinsics along with depth and pose, by learning from monocular image sequences alone. In addition, by utilizing efficient sub-pixel convolutions, we show that high fidelity depth estimates can be obtained. We also embed pixel-wise uncertainty estimation into the framework, to emphasize the possible applicability of this work in practical domain. Finally, we demonstrate the possibility of accurate prediction of depth information without prior knowledge of camera intrinsics, while outperforming the existing state-of-the-art approaches on KITTI benchmark.
In this paper, we propose a novel method, X-Distill, to improve the self-supervised training of monocular depth via cross-task knowledge distillation from semantic segmentation to depth estimation. More specifically, during training, we utilize a pretrained semantic segmentation teacher network and transfer its semantic knowledge to the depth network. In order to enable such knowledge distillation across two different visual tasks, we introduce a small, trainable network that translates the predicted depth map to a semantic segmentation map, which can then be supervised by the teacher network. In this way, this small network enables the backpropagation from the semantic segmentation teacher's supervision to the depth network during training. In addition, since the commonly used object classes in semantic segmentation are not directly transferable to depth, we study the visual and geometric characteristics of the objects and design a new way of grouping them that can be shared by both tasks. It is noteworthy that our approach only modifies the training process and does not incur additional computation during inference. We extensively evaluate the efficacy of our proposed approach on the standard KITTI benchmark and compare it with the latest state of the art. We further test the generalizability of our approach on Make3D. Overall, the results show that our approach significantly improves the depth estimation accuracy and outperforms the state of the art.
Despite recent improvement of supervised monocular depth estimation, the lack of high quality pixel-wise ground truth annotations has become a major hurdle for further progress. In this work, we propose a new unsupervised depth estimation method based on pseudo supervision mechanism by training a teacher-student network with knowledge distillation. It strategically integrates the advantages of supervised and unsupervised monocular depth estimation, as well as unsupervised binocular depth estimation. Specifically, the teacher network takes advantage of the effectiveness of binocular depth estimation to produce accurate disparity maps, which are then used as the pseudo ground truth to train the student network for monocular depth estimation. This effectively converts the problem of unsupervised learning to supervised learning. Our extensive experimental results demonstrate that the proposed method outperforms the state-of-the-art on the KITTI benchmark.
Self-supervised monocular depth estimation approaches either ignore independently moving objects in the scene or need a separate segmentation step to identify them. We propose MonoDepthSeg to jointly estimate depth and segment moving objects from monocular video without using any ground-truth labels. We decompose the scene into a fixed number of components where each component corresponds to a region on the image with its own transformation matrix representing its motion. We estimate both the mask and the motion of each component efficiently with a shared encoder. We evaluate our method on three driving datasets and show that our model clearly improves depth estimation while decomposing the scene into separately moving components.