Purpose: Stereo matching methods that enable depth estimation are crucial for visualization enhancement applications in computer-assisted surgery (CAS). Learning-based stereo matching methods are promising to predict accurate results on laparoscopic images. However, they require a large amount of training data, and their performance may be degraded due to domain shifts. Methods: Maintaining robustness and improving the accuracy of learning-based methods are still open problems. To overcome the limitations of learning-based methods, we propose a disparity refinement framework consisting of a local disparity refinement method and a global disparity refinement method to improve the results of learning-based stereo matching methods in a cross-domain setting. Those learning-based stereo matching methods are pre-trained on a large public dataset of natural images and are tested on two datasets of laparoscopic images. Results: Qualitative and quantitative results suggest that our proposed disparity framework can effectively refine disparity maps when they are noise-corrupted on an unseen dataset, without compromising prediction accuracy when the network can generalize well on an unseen dataset. Conclusion: Our proposed disparity refinement framework could work with learning-based methods to achieve robust and accurate disparity prediction. Yet, as a large laparoscopic dataset for training learning-based methods does not exist and the generalization ability of networks remains to be improved, the incorporation of the proposed disparity refinement framework into existing networks will contribute to improving their overall accuracy and robustness associated with depth estimation.
Multiview depth imagery will play a critical role in free-viewpoint television. This technology requires high quality virtual view synthesis to enable viewers to move freely in a dynamic real world scene. Depth imagery at different viewpoints is used to synthesize an arbitrary number of novel views. Usually, depth images at multiple viewpoints are estimated individually by stereo-matching algorithms, and hence, show lack of interview consistency. This inconsistency affects the quality of view synthesis negatively. This paper proposes a method for depth consistency testing in depth difference subspace to enhance the depth representation of a scene across multiple viewpoints. Furthermore, we propose a view synthesis algorithm that uses the obtained consistency information to improve the visual quality of virtual views at arbitrary viewpoints. Our method helps us to find a linear subspace for our depth difference measurements in which we can test the inter-view consistency efficiently. With this, our approach is able to enhance the depth information for real world scenes. In combination with our consistency-adaptive view synthesis, we improve the visual experience of the free-viewpoint user. The experiments show that our approach enhances the objective quality of virtual views by up to 1.4 dB. The advantage for the subjective quality is also demonstrated.
In this paper, we propose CGI-Stereo, a novel neural network architecture that can concurrently achieve real-time performance, competitive accuracy, and strong generalization ability. The core of our CGI-Stereo is a Context and Geometry Fusion (CGF) block which adaptively fuses context and geometry information for more effective cost aggregation and meanwhile provides feedback to feature learning to guide more effective contextual feature extraction. The proposed CGF can be easily embedded into many existing stereo matching networks, such as PSMNet, GwcNet and ACVNet. The resulting networks show a significant improvement in accuracy. Specially, the model which incorporates our CGF with ACVNet ranks $1^{st}$ on the KITTI 2012 and 2015 leaderboards among all the published methods. We further propose an informative and concise cost volume, named Attention Feature Volume (AFV), which exploits a correlation volume as attention weights to filter a feature volume. Based on CGF and AFV, the proposed CGI-Stereo outperforms all other published real-time methods on KITTI benchmarks and shows better generalization ability than other real-time methods. Code is available at https://github.com/gangweiX/CGI-Stereo.
Several leading methods on public benchmarks for depth-from-stereo rely on memory-demanding 4D cost volumes and computationally intensive 3D convolutions for feature matching. We suggest a new way to process the 4D cost volume where we merge two different concepts in one deeply integrated framework to achieve a symbiotic relationship. A feature matching part is responsible for identifying matching pixels pairs along the baseline while a concurrent image volume part is inspired by depth-from-mono CNNs. However, instead of predicting depth directly from image features, it provides additional context to resolve ambiguities during pixel matching. More technically, the processing of the 4D cost volume is separated into a 2D propagation and a 3D propagation part. Starting from feature maps of the left image, the 2D propagation assists the 3D propagation part of the cost volume at different layers by adding visual features to the geometric context. By combining both parts, we can safely reduce the scale of 3D convolution layers in the matching part without sacrificing accuracy. Experiments demonstrate that our end-to-end trained CNN is ranked 2nd on KITTI2012 and ETH3D benchmarks while being significantly faster than the 1st-ranked method. Furthermore, we notice that the coupling of image and matching-volume improves fine-scale details as demonstrated by our qualitative analysis.
In this paper, we propose a low error rate and real-time stereo vision system on GPU. Many stereo vision systems on GPU have been proposed to date. In those systems, the error rates and the processing speed are in trade-off relationship. We propose a real-time stereo vision system on GPU for the high resolution images. This system also maintains a low error rate compared to other fast systems. In our approach, we have implemented the cost aggregation (CA), cross-checking and median filter on GPU in order to realize the real-time processing. Its processing speed is 40 fps for 1436x992 pixels images when the maximum disparity is 145, and its error rate is the lowest among the GPU systems which are faster than 30 fps.
Mobile stereo-matching systems have become an important part of many applications, such as automated-driving vehicles and autonomous robots. Accurate stereo-matching methods usually lead to high computational complexity; however, mobile platforms have only limited hardware resources to keep their power consumption low; this makes it difficult to maintain both an acceptable processing speed and accuracy on mobile platforms. To resolve this trade-off, we herein propose a novel acceleration approach for the well-known zero-means normalized cross correlation (ZNCC) matching cost calculation algorithm on a Jetson Tx2 embedded GPU. In our method for accelerating ZNCC, target images are scanned in a zigzag fashion to efficiently reuse one pixel's computation for its neighboring pixels; this reduces the amount of data transmission and increases the utilization of on-chip registers, thus increasing the processing speed. As a result, our method is 2X faster than the traditional image scanning method, and 26% faster than the latest NCC method. By combining this technique with the domain transformation (DT) algorithm, our system show real-time processing speed of 32 fps, on a Jetson Tx2 GPU for 1,280x384 pixel images with a maximum disparity of 128. Additionally, the evaluation results on the KITTI 2015 benchmark show that our combined system is more accurate than the same algorithm combined with census by 7.26%, while maintaining almost the same processing speed.
We present TemporalStereo, a coarse-to-fine stereo matching network that is highly efficient, and able to effectively exploit the past geometry and context information to boost matching accuracy. Our network leverages sparse cost volume and proves to be effective when a single stereo pair is given. However, its peculiar ability to use spatio-temporal information across stereo sequences allows TemporalStereo to alleviate problems such as occlusions and reflective regions while enjoying high efficiency also in this latter case. Notably, our model -- trained once with stereo videos -- can run in both single-pair and temporal modes seamlessly. Experiments show that our network relying on camera motion is robust even to dynamic objects when running on videos. We validate TemporalStereo through extensive experiments on synthetic (SceneFlow, TartanAir) and real (KITTI 2012, KITTI 2015) datasets. Our model achieves state-of-the-art performance on any of these datasets. Code is available at \url{https://github.com/youmi-zym/TemporalStereo.git}.
Despite impressive performance for high-level downstream tasks, self-supervised pre-training methods have not yet fully delivered on dense geometric vision tasks such as stereo matching or optical flow. The application of self-supervised concepts, such as instance discrimination or masked image modeling, to geometric tasks is an active area of research. In this work, we build on the recent cross-view completion framework, a variation of masked image modeling that leverages a second view from the same scene which makes it well suited for binocular downstream tasks. The applicability of this concept has so far been limited in at least two ways: (a) by the difficulty of collecting real-world image pairs -- in practice only synthetic data have been used -- and (b) by the lack of generalization of vanilla transformers to dense downstream tasks for which relative position is more meaningful than absolute position. We explore three avenues of improvement. First, we introduce a method to collect suitable real-world image pairs at large scale. Second, we experiment with relative positional embeddings and show that they enable vision transformers to perform substantially better. Third, we scale up vision transformer based cross-completion architectures, which is made possible by the use of large amounts of data. With these improvements, we show for the first time that state-of-the-art results on stereo matching and optical flow can be reached without using any classical task-specific techniques like correlation volume, iterative estimation, image warping or multi-scale reasoning, thus paving the way towards universal vision models.
We present a unified formulation and model for three motion and 3D perception tasks: optical flow, rectified stereo matching and unrectified stereo depth estimation from posed images. Unlike previous specialized architectures for each specific task, we formulate all three tasks as a unified dense correspondence matching problem, which can be solved with a single model by directly comparing feature similarities. Such a formulation calls for discriminative feature representations, which we achieve using a Transformer, in particular the cross-attention mechanism. We demonstrate that cross-attention enables integration of knowledge from another image via cross-view interactions, which greatly improves the quality of the extracted features. Our unified model naturally enables cross-task transfer since the model architecture and parameters are shared across tasks. We outperform RAFT with our unified model on the challenging Sintel dataset, and our final model that uses a few additional task-specific refinement steps outperforms or compares favorably to recent state-of-the-art methods on 10 popular flow, stereo and depth datasets, while being simpler and more efficient in terms of model design and inference speed.
Event cameras are novel bio-inspired vision sensors that output pixel-level intensity changes in microsecond accuracy with a high dynamic range and low power consumption. Despite these advantages, event cameras cannot be directly applied to computational imaging tasks due to the inability to obtain high-quality intensity and events simultaneously. This paper aims to connect a standalone event camera and a modern intensity camera so that the applications can take advantage of both two sensors. We establish this connection through a multi-modal stereo matching task. We first convert events to a reconstructed image and extend the existing stereo networks to this multi-modality condition. We propose a self-supervised method to train the multi-modal stereo network without using ground truth disparity data. The structure loss calculated on image gradients is used to enable self-supervised learning on such multi-modal data. Exploiting the internal stereo constraint between views with different modalities, we introduce general stereo loss functions, including disparity cross-consistency loss and internal disparity loss, leading to improved performance and robustness compared to existing approaches. The experiments demonstrate the effectiveness of the proposed method, especially the proposed general stereo loss functions, on both synthetic and real datasets. At last, we shed light on employing the aligned events and intensity images in downstream tasks, e.g., video interpolation application.
We introduce visual hints expansion for guiding stereo matching to improve generalization. Our work is motivated by the robustness of Visual Inertial Odometry (VIO) in computer vision and robotics, where a sparse and unevenly distributed set of feature points characterizes a scene. To improve stereo matching, we propose to elevate 2D hints to 3D points. These sparse and unevenly distributed 3D visual hints are expanded using a 3D random geometric graph, which enhances the learning and inference process. We evaluate our proposal on multiple widely adopted benchmarks and show improved performance without access to additional sensors other than the image sequence. To highlight practical applicability and symbiosis with visual odometry, we demonstrate how our methods run on embedded hardware.
Stereo Matching is one of the classical problems in computer vision for the extraction of 3D information but still controversial for accuracy and processing costs. The use of matching techniques and cost functions is crucial in the development of the disparity map. This paper presents a comparative study of six different stereo matching algorithms including Block Matching (BM), Block Matching with Dynamic Programming (BMDP), Belief Propagation (BP), Gradient Feature Matching (GF), Histogram of Oriented Gradient (HOG), and the proposed method. Also three cost functions namely Mean Squared Error (MSE), Sum of Absolute Differences (SAD), Normalized Cross-Correlation (NCC) were used and compared. The stereo images used in this study were from the Middlebury Stereo Datasets provided with perfect and imperfect calibrations. Results show that the selection of matching function is quite important and also depends on the images properties. Results showed that the BP algorithm in most cases provided better results getting accuracies over 95%.
Stereo correspondence matching is an essential part of the multi-step stereo depth estimation process. This paper revisits the depth estimation problem, avoiding the explicit stereo matching step using a simple two-tower convolutional neural network. The proposed algorithm is entitled as 2T-UNet. The idea behind 2T-UNet is to replace cost volume construction with twin convolution towers. These towers have an allowance for different weights between them. Additionally, the input for twin encoders in 2T-UNet are different compared to the existing stereo methods. Generally, a stereo network takes a right and left image pair as input to determine the scene geometry. However, in the 2T-UNet model, the right stereo image is taken as one input and the left stereo image along with its monocular depth clue information, is taken as the other input. Depth clues provide complementary suggestions that help enhance the quality of predicted scene geometry. The 2T-UNet surpasses state-of-the-art monocular and stereo depth estimation methods on the challenging Scene flow dataset, both quantitatively and qualitatively. The architecture performs incredibly well on complex natural scenes, highlighting its usefulness for various real-time applications. Pretrained weights and code will be made readily available.
Deep learning (DL) stereo matching methods gained great attention in remote sensing satellite datasets. However, most of these existing studies conclude assessments based only on a few/single stereo images lacking a systematic evaluation on how robust DL methods are on satellite stereo images with varying radiometric and geometric configurations. This paper provides an evaluation of four DL stereo matching methods through hundreds of multi-date multi-site satellite stereo pairs with varying geometric configurations, against the traditional well-practiced Census-SGM (Semi-global matching), to comprehensively understand their accuracy, robustness, generalization capabilities, and their practical potential. The DL methods include a learning-based cost metric through convolutional neural networks (MC-CNN) followed by SGM, and three end-to-end (E2E) learning models using Geometry and Context Network (GCNet), Pyramid Stereo Matching Network (PSMNet), and LEAStereo. Our experiments show that E2E algorithms can achieve upper limits of geometric accuracies, while may not generalize well for unseen data. The learning-based cost metric and Census-SGM are rather robust and can consistently achieve acceptable results. All DL algorithms are robust to geometric configurations of stereo pairs and are less sensitive in comparison to the Census-SGM, while learning-based cost metrics can generalize on satellite images when trained on different datasets (airborne or ground-view).
Stereo-matching is a fundamental problem in computer vision. Despite recent progress by deep learning, improving the robustness is ineluctable when deploying stereo-matching models to real-world applications. Different from the common practices, i.e., developing an elaborate model to achieve robustness, we argue that collecting multiple available datasets for training is a cheaper way to increase generalization ability. Specifically, this report presents an improved RaftStereo trained with a mixed dataset of seven public datasets for the robust vision challenge (denoted as iRaftStereo_RVC). When evaluated on the training sets of Middlebury, KITTI-2015, and ETH3D, the model outperforms its counterparts trained with only one dataset, such as the popular Sceneflow. After fine-tuning the pre-trained model on the three datasets of the challenge, it ranks at 2nd place on the stereo leaderboard, demonstrating the benefits of mixed dataset pre-training.
Stereo depth estimation is of great interest for computer vision research. However, existing methods struggles to generalize and predict reliably in hazardous regions, such as large uniform regions. To overcome these limitations, we propose Context Enhanced Path (CEP). CEP improves the generalization and robustness against common failure cases in existing solutions by capturing the long-range global information. We construct our stereo depth estimation model, Context Enhanced Stereo Transformer (CSTR), by plugging CEP into the state-of-the-art stereo depth estimation method Stereo Transformer. CSTR is examined on distinct public datasets, such as Scene Flow, Middlebury-2014, KITTI-2015, and MPI-Sintel. We find CSTR outperforms prior approaches by a large margin. For example, in the zero-shot synthetic-to-real setting, CSTR outperforms the best competing approaches on Middlebury-2014 dataset by 11%. Our extensive experiments demonstrate that the long-range information is critical for stereo matching task and CEP successfully captures such information.
While recent camera-only 3D detection methods leverage multiple timesteps, the limited history they use significantly hampers the extent to which temporal fusion can improve object perception. Observing that existing works' fusion of multi-frame images are instances of temporal stereo matching, we find that performance is hindered by the interplay between 1) the low granularity of matching resolution and 2) the sub-optimal multi-view setup produced by limited history usage. Our theoretical and empirical analysis demonstrates that the optimal temporal difference between views varies significantly for different pixels and depths, making it necessary to fuse many timesteps over long-term history. Building on our investigation, we propose to generate a cost volume from a long history of image observations, compensating for the coarse but efficient matching resolution with a more optimal multi-view matching setup. Further, we augment the per-frame monocular depth predictions used for long-term, coarse matching with short-term, fine-grained matching and find that long and short term temporal fusion are highly complementary. While maintaining high efficiency, our framework sets new state-of-the-art on nuScenes, achieving first place on the test set and outperforming previous best art by 5.2% mAP and 3.7% NDS on the validation set. Code will be released $\href{https://github.com/Divadi/SOLOFusion}{here.}$
We propose a non-learning depth completion method for a sparse depth map captured using a light detection and ranging (LiDAR) sensor guided by a pair of stereo images. Generally, conventional stereo-aided depth completion methods have two limiations. (i) They assume the given sparse depth map is accurately aligned to the input image, whereas the alignment is difficult to achieve in practice. (ii) They have limited accuracy in the long range because the depth is estimated by pixel disparity. To solve the abovementioned limitations, we propose selective stereo matching (SSM) that searches the most appropriate depth value for each image pixel from its neighborly projected LiDAR points based on an energy minimization framework. This depth selection approach can handle any type of mis-projection. Moreover, SSM has an advantage in terms of long-range depth accuracy because it directly uses the LiDAR measurement rather than the depth acquired from the stereo. SSM is a discrete process; thus, we apply variational smoothing with binary anisotropic diffusion tensor (B-ADT) to generate a continuous depth map while preserving depth discontinuity across object boundaries. Experimentally, compared with the previous state-of-the-art stereo-aided depth completion, the proposed method reduced the mean absolute error (MAE) of the depth estimation to 0.65 times and demonstrated approximately twice more accurate estimation in the long range. Moreover, under various LiDAR-camera calibration errors, the proposed method reduced the depth estimation MAE to 0.34-0.93 times from previous depth completion methods.
Stereo matching is a fundamental building block for many vision and robotics applications. An informative and concise cost volume representation is vital for stereo matching of high accuracy and efficiency. In this paper, we present a novel cost volume construction method, named attention concatenation volume (ACV), which generates attention weights from correlation clues to suppress redundant information and enhance matching-related information in the concatenation volume. The ACV can be seamlessly embedded into most stereo matching networks, the resulting networks can use a more lightweight aggregation network and meanwhile achieve higher accuracy. We further design a fast version of ACV to enable real-time performance, named Fast-ACV, which generates high likelihood disparity hypotheses and the corresponding attention weights from low-resolution correlation clues to significantly reduce computational and memory cost and meanwhile maintain a satisfactory accuracy. The core idea of our Fast-ACV is volume attention propagation (VAP) which can automatically select accurate correlation values from an upsampled correlation volume and propagate these accurate values to the surroundings pixels with ambiguous correlation clues. Furthermore, we design a highly accurate network ACVNet and a real-time network Fast-ACVNet based on our ACV and Fast-ACV respectively, which achieve the state-of-the-art performance on several benchmarks (i.e., our ACVNet ranks the 2nd on KITTI 2015 and Scene Flow, and the 3rd on KITTI 2012 and ETH3D among all the published methods; our Fast-ACVNet outperforms almost all state-of-the-art real-time methods on Scene Flow, KITTI 2012 and 2015 and meanwhile has better generalization ability)
Obstacle detection is a safety-critical problem in robot navigation, where stereo matching is a popular vision-based approach. While deep neural networks have shown impressive results in computer vision, most of the previous obstacle detection works only leverage traditional stereo matching techniques to meet the computational constraints for real-time feedback. This paper proposes a computationally efficient method that employs a deep neural network to detect occupancy from stereo images directly. Instead of learning the point cloud correspondence from the stereo data, our approach extracts the compact obstacle distribution based on volumetric representations. In addition, we prune the computation of safety irrelevant spaces in a coarse-to-fine manner based on octrees generated by the decoder. As a result, we achieve real-time performance on the onboard computer (NVIDIA Jetson TX2). Our approach detects obstacles accurately in the range of 32 meters and achieves better IoU (Intersection over Union) and CD (Chamfer Distance) scores with only 2% of the computation cost of the state-of-the-art stereo model. Furthermore, we validate our method's robustness and real-world feasibility through autonomous navigation experiments with a real robot. Hence, our work contributes toward closing the gap between the stereo-based system in robot perception and state-of-the-art stereo models in computer vision. To counter the scarcity of high-quality real-world indoor stereo datasets, we collect a 1.36 hours stereo dataset with a mobile robot which is used to fine-tune our model. The dataset, the code, and further details including additional visualizations are available at https://lhy.xyz/stereovoxelnet