We introduce TAPIP3D, a novel approach for long-term 3D point tracking in monocular RGB and RGB-D videos. TAPIP3D represents videos as camera-stabilized spatio-temporal feature clouds, leveraging depth and camera motion information to lift 2D video features into a 3D world space where camera motion is effectively canceled. TAPIP3D iteratively refines multi-frame 3D motion estimates within this stabilized representation, enabling robust tracking over extended periods. To manage the inherent irregularities of 3D point distributions, we propose a Local Pair Attention mechanism. This 3D contextualization strategy effectively exploits spatial relationships in 3D, forming informative feature neighborhoods for precise 3D trajectory estimation. Our 3D-centric approach significantly outperforms existing 3D point tracking methods and even enhances 2D tracking accuracy compared to conventional 2D pixel trackers when accurate depth is available. It supports inference in both camera coordinates (i.e., unstabilized) and world coordinates, and our results demonstrate that compensating for camera motion improves tracking performance. Our approach replaces the conventional 2D square correlation neighborhoods used in prior 2D and 3D trackers, leading to more robust and accurate results across various 3D point tracking benchmarks. Project Page: https://tapip3d.github.io
Modern mobile burst photography pipelines capture and merge a short sequence of frames to recover an enhanced image, but often disregard the 3D nature of the scene they capture, treating pixel motion between images as a 2D aggregation problem. We show that in a ''long-burst'', forty-two 12-megapixel RAW frames captured in a two-second sequence, there is enough parallax information from natural hand tremor alone to recover high-quality scene depth. To this end, we devise a test-time optimization approach that fits a neural RGB-D representation to long-burst data and simultaneously estimates scene depth and camera motion. Our plane plus depth model is trained end-to-end, and performs coarse-to-fine refinement by controlling which multi-resolution volume features the network has access to at what time during training. We validate the method experimentally, and demonstrate geometrically accurate depth reconstructions with no additional hardware or separate data pre-processing and pose-estimation steps.
In autonomous driving, monocular sequences contain lots of information. Monocular depth estimation, camera ego-motion estimation and optical flow estimation in consecutive frames are high-profile concerns recently. By analyzing tasks above, pixels in the middle frame are modeled into three parts: the rigid region, the non-rigid region, and the occluded region. In joint unsupervised training of depth and pose, we can segment the occluded region explicitly. The occlusion information is used in unsupervised learning of depth, pose and optical flow, as the image reconstructed by depth-pose and optical flow will be invalid in occluded regions. A less-than-mean mask is designed to further exclude the mismatched pixels interfered with by motion or illumination change in the training of depth and pose networks. This method is also used to exclude some trivial mismatched pixels in the training of the optical flow network. Maximum normalization is proposed for depth smoothness term to restrain depth degradation in textureless regions. In the occluded region, as depth and camera motion can provide more reliable motion estimation, they can be used to instruct unsupervised learning of optical flow. Our experiments in KITTI dataset demonstrate that the model based on three regions, full and explicit segmentation of the occlusion region, the rigid region, and the non-rigid region with corresponding unsupervised losses can improve performance on three tasks significantly. The source code is available at: https://github.com/guangmingw/DOPlearning.
We consider the problem of unsupervised camera pose estimation. Given an input video sequence, our goal is to estimate the camera pose (i.e. the camera motion) between consecutive frames. Traditionally, this problem is tackled by placing strict constraints on the transformation vector or by incorporating optical flow through a complex pipeline. We propose an alternative approach that utilizes a compositional re-estimation process for camera pose estimation. Given an input, we first estimate a depth map. Our method then iteratively estimates the camera motion based on the estimated depth map. Our approach significantly improves the predicted camera motion both quantitatively and visually. Furthermore, the re-estimation resolves the problem of out-of-boundaries pixels in a novel and simple way. Another advantage of our approach is that it is adaptable to other camera pose estimation approaches. Experimental analysis on KITTI benchmark dataset demonstrates that our method outperforms existing state-of-the-art approaches in unsupervised camera ego-motion estimation.
Learning to predict scene depth and camera motion from RGB inputs only is a challenging task. Most existing learning based methods deal with this task in a supervised manner which require ground-truth data that is expensive to acquire. More recent approaches explore the possibility of estimating scene depth and camera pose in a self-supervised learning framework. Despite encouraging results are shown, current methods either learn from monocular videos for depth and pose and typically do so without enforcing multi-view geometry constraints between scene structure and camera motion, or require stereo sequences as input where the ground-truth between-frame motion parameters need to be known. In this paper we propose to jointly optimize the scene depth and camera motion via incorporating differentiable Bundle Adjustment (BA) layer by minimizing the feature-metric error, and then form the photometric consistency loss with view synthesis as the final supervisory signal. The proposed approach only needs unlabeled monocular videos as input, and extensive experiments on the KITTI and Cityscapes dataset show that our method achieves state-of-the-art results in self-supervised approaches using monocular videos as input, and even gains advantage to the line of methods that learns from calibrated stereo sequences (i.e. with pose supervision).
We present a method for decomposing the 3D scene flow observed from a moving stereo rig into stationary scene elements and dynamic object motion. Our unsupervised learning framework jointly reasons about the camera motion, optical flow, and 3D motion of moving objects. Three cooperating networks predict stereo matching, camera motion, and residual flow, which represents the flow component due to object motion and not from camera motion. Based on rigid projective geometry, the estimated stereo depth is used to guide the camera motion estimation, and the depth and camera motion are used to guide the residual flow estimation. We also explicitly estimate the 3D scene flow of dynamic objects based on the residual flow and scene depth. Experiments on the KITTI dataset demonstrate the effectiveness of our approach and show that our method outperforms other state-of-the-art algorithms on the optical flow and visual odometry tasks.
Unsupervised learning based depth estimation methods have received more and more attention as they do not need vast quantities of densely labeled data for training which are touch to acquire. In this paper, we propose a novel unsupervised monocular video depth estimation method in natural scenes by taking advantage of the state-of-the-art method of Zhou et al. which jointly estimates depth and camera motion. Our method advances beyond the baseline method by three aspects: 1) we add an additional signal as supervision to the baseline method by incorporating left-right binocular images reconstruction loss based on the estimated disparities, thus the left frame can be reconstructed by the temporal frames and right frames of stereo vision; 2) the network is trained by jointly using two kinds of view syntheses loss and left-right disparity consistency regularization to estimate depth and pose simultaneously; 3) we use the edge aware smooth L2 regularization to smooth the depth map while preserving the contour of the target. Extensive experiments on the KITTI autonomous driving dataset and Make3D dataset indicate the superiority of our algorithm in training efficiency. We can achieve competitive results with the baseline by only 3/5 times training data. The experimental results also show that our method even outperforms the classical supervised methods that using either ground truth depth or given pose for training.
Single view depth estimation models can be trained from video footage using a self-supervised end-to-end approach with view synthesis as the supervisory signal. This is achieved with a framework that predicts depth and camera motion, with a loss based on reconstructing a target video frame from temporally adjacent frames. In this context, occlusion relates to parts of a scene that can be observed in the target frame but not in a frame used for image reconstruction. Since the image reconstruction is based on sampling from the adjacent frame, and occluded areas by definition cannot be sampled, reconstructed occluded areas corrupt to the supervisory signal. In previous work arXiv:1806.01260 occlusion is handled based on reconstruction error; at each pixel location, only the reconstruction with the lowest error is included in the loss. The current study aims to determine whether performance improvements of depth estimation models can be gained by during training only ignoring those regions that are affected by occlusion. In this work we introduce occlusion mask, a mask that during training can be used to specifically ignore regions that cannot be reconstructed due to occlusions. Occlusion mask is based entirely on predicted depth information. We introduce two novel loss formulations which incorporate the occlusion mask. The method and implementation of arXiv:1806.01260 serves as the foundation for our modifications as well as the baseline in our experiments. We demonstrate that (i) incorporating occlusion mask in the loss function improves the performance of single image depth prediction models on the KITTI benchmark. (ii) loss functions that select from reconstructions based on error are able to ignore some of the reprojection error caused by object motion.
Recent work has shown that CNN-based depth and ego-motion estimators can be learned using unlabelled monocular videos. However, the performance is limited by unidentified moving objects that violate the underlying static scene assumption in geometric image reconstruction. More significantly, due to lack of proper constraints, networks output scale-inconsistent results over different samples, i.e., the ego-motion network cannot provide full camera trajectories over a long video sequence because of the per-frame scale ambiguity. This paper tackles these challenges by proposing a geometry consistency loss for scale-consistent predictions and an induced self-discovered mask for handling moving objects and occlusions. Since we do not leverage multi-task learning like recent works, our framework is much simpler and more efficient. Comprehensive evaluation results demonstrate that our depth estimator achieves the state-of-the-art performance on the KITTI dataset. Moreover, we show that our ego-motion network is able to predict a globally scale-consistent camera trajectory for long video sequences, and the resulting visual odometry accuracy is competitive with the recent model that is trained using stereo videos. To the best of our knowledge, this is the first work to show that deep networks trained using unlabelled monocular videos can predict globally scale-consistent camera trajectories over a long video sequence.
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).
We present an approach which takes advantage of both structure and semantics for unsupervised monocular learning of depth and ego-motion. More specifically, we model the motion of individual objects and learn their 3D motion vector jointly with depth and ego-motion. We obtain more accurate results, especially for challenging dynamic scenes not addressed by previous approaches. This is an extended version of Casser et al. [AAAI'19]. Code and models have been open sourced at https://sites.google.com/corp/view/struct2depth.
Dynamic scenes that contain both object motion and egomotion are a challenge for monocular visual odometry (VO). Another issue with monocular VO is the scale ambiguity, i.e. these methods cannot estimate scene depth and camera motion in real scale. Here, we propose a learning based approach to predict camera motion parameters directly from optic flow, by marginalizing depthmap variations and outliers. This is achieved by learning a sparse overcomplete basis set of egomotion in an autoencoder network, which is able to eliminate irrelevant components of optic flow for the task of camera parameter or motionfield estimation. The model is trained using a sparsity regularizer and a supervised egomotion loss, and achieves the state-of-the-art performances on trajectory prediction and camera rotation prediction tasks on KITTI and Virtual KITTI datasets, respectively. The sparse latent space egomotion representation learned by the model is robust and requires only 5% of the hidden layer neurons to maintain the best trajectory prediction accuracy on KITTI dataset. Additionally, in presence of depth information, the proposed method demonstrates faithful object velocity prediction for wide range of object sizes and speeds by global compensation of predicted egomotion and a divisive normalization procedure.
The RGB-D camera maintains a limited range for working and is hard to accurately measure the depth information in a far distance. Besides, the RGB-D camera will easily be influenced by strong lighting and other external factors, which will lead to a poor accuracy on the acquired environmental depth information. Recently, deep learning technologies have achieved great success in the visual SLAM area, which can directly learn high-level features from the visual inputs and improve the estimation accuracy of the depth information. Therefore, deep learning technologies maintain the potential to extend the source of the depth information and improve the performance of the SLAM system. However, the existing deep learning-based methods are mainly supervised and require a large amount of ground-truth depth data, which is hard to acquire because of the realistic constraints. In this paper, we first present an unsupervised learning framework, which not only uses image reconstruction for supervising but also exploits the pose estimation method to enhance the supervised signal and add training constraints for the task of monocular depth and camera motion estimation. Furthermore, we successfully exploit our unsupervised learning framework to assist the traditional ORB-SLAM system when the initialization module of ORB-SLAM method could not match enough features. Qualitative and quantitative experiments have shown that our unsupervised learning framework performs the depth estimation task comparable to the supervised methods and outperforms the previous state-of-the-art approach by $13.5\%$ on KITTI dataset. Besides, our unsupervised learning framework could significantly accelerate the initialization process of ORB-SLAM system and effectively improve the accuracy on environmental mapping in strong lighting and weak texture scenes.
A reliable sense-and-avoid system is critical to enabling safe autonomous operation of unmanned aircraft. Existing sense-and-avoid methods often require specialized sensors that are too large or power intensive for use on small unmanned vehicles. This paper presents a method to estimate object distances based on visual image sequences, allowing for the use of low-cost, on-board monocular cameras as simple collision avoidance sensors. We present a deep recurrent convolutional neural network and training method to generate depth maps from video sequences. Our network is trained using simulated camera and depth data generated with Microsoft's AirSim simulator. Empirically, we show that our model achieves superior performance compared to models generated using prior methods.We further demonstrate that the method can be used for sense-and-avoid of obstacles in simulation.
As 360{\deg} cameras become prevalent in many autonomous systems (e.g., self-driving cars and drones), efficient 360{\deg} perception becomes more and more important. We propose a novel self-supervised learning approach for predicting the omnidirectional depth and camera motion from a 360{\deg} video. In particular, starting from the SfMLearner, which is designed for cameras with normal field-of-view, we introduce three key features to process 360{\deg} images efficiently. Firstly, we convert each image from equirectangular projection to cubic projection in order to avoid image distortion. In each network layer, we use Cube Padding (CP), which pads intermediate features from adjacent faces, to avoid image boundaries. Secondly, we propose a novel "spherical" photometric consistency constraint on the whole viewing sphere. In this way, no pixel will be projected outside the image boundary which typically happens in images with normal field-of-view. Finally, rather than naively estimating six independent camera motions (i.e., naively applying SfM-Learner to each face on a cube), we propose a novel camera pose consistency loss to ensure the estimated camera motions reaching consensus. To train and evaluate our approach, we collect a new PanoSUNCG dataset containing a large amount of 360{\deg} videos with groundtruth depth and camera motion. Our approach achieves state-of-the-art depth prediction and camera motion estimation on PanoSUNCG with faster inference speed comparing to equirectangular. In real-world indoor videos, our approach can also achieve qualitatively reasonable depth prediction by acquiring model pre-trained on PanoSUNCG.
We present an unsupervised learning framework for simultaneously training single-view depth prediction and optical flow estimation models using unlabeled video sequences. Existing unsupervised methods often exploit brightness constancy and spatial smoothness priors to train depth or flow models. In this paper, we propose to leverage geometric consistency as additional supervisory signals. Our core idea is that for rigid regions we can use the predicted scene depth and camera motion to synthesize 2D optical flow by backprojecting the induced 3D scene flow. The discrepancy between the rigid flow (from depth prediction and camera motion) and the estimated flow (from optical flow model) allows us to impose a cross-task consistency loss. While all the networks are jointly optimized during training, they can be applied independently at test time. Extensive experiments demonstrate that our depth and flow models compare favorably with state-of-the-art unsupervised methods.
Learning to estimate 3D geometry in a single image by watching unlabeled videos via deep convolutional network has made significant process recently. Current state-of-the-art (SOTA) methods, are based on the learning framework of rigid structure-from-motion, where only 3D camera ego motion is modeled for geometry estimation.However, moving objects also exist in many videos, e.g. moving cars in a street scene. In this paper, we tackle such motion by additionally incorporating per-pixel 3D object motion into the learning framework, which provides holistic 3D scene flow understanding and helps single image geometry estimation. Specifically, given two consecutive frames from a video, we adopt a motion network to predict their relative 3D camera pose and a segmentation mask distinguishing moving objects and rigid background. An optical flow network is used to estimate dense 2D per-pixel correspondence. A single image depth network predicts depth maps for both images. The four types of information, i.e. 2D flow, camera pose, segment mask and depth maps, are integrated into a differentiable holistic 3D motion parser (HMP), where per-pixel 3D motion for rigid background and moving objects are recovered. We design various losses w.r.t. the two types of 3D motions for training the depth and motion networks, yielding further error reduction for estimated geometry. Finally, in order to solve the 3D motion confusion from monocular videos, we combine stereo images into joint training. Experiments on KITTI 2015 dataset show that our estimated geometry, 3D motion and moving object masks, not only are constrained to be consistent, but also significantly outperforms other SOTA algorithms, demonstrating the benefits of our approach.
In this paper we formulate structure from motion as a learning problem. We train a convolutional network end-to-end to compute depth and camera motion from successive, unconstrained image pairs. The architecture is composed of multiple stacked encoder-decoder networks, the core part being an iterative network that is able to improve its own predictions. The network estimates not only depth and motion, but additionally surface normals, optical flow between the images and confidence of the matching. A crucial component of the approach is a training loss based on spatial relative differences. Compared to traditional two-frame structure from motion methods, results are more accurate and more robust. In contrast to the popular depth-from-single-image networks, DeMoN learns the concept of matching and, thus, better generalizes to structures not seen during training.
We present an unsupervised learning framework for the task of monocular depth and camera motion estimation from unstructured video sequences. We achieve this by simultaneously training depth and camera pose estimation networks using the task of view synthesis as the supervisory signal. The networks are thus coupled via the view synthesis objective during training, but can be applied independently at test time. Empirical evaluation on the KITTI dataset demonstrates the effectiveness of our approach: 1) monocular depth performing comparably with supervised methods that use either ground-truth pose or depth for training, and 2) pose estimation performing favorably with established SLAM systems under comparable input settings.