SRH-Net: Stacked Recurrent Hourglass Network for Stereo Matching

Du, Hongzhi; Li, Yanyan; Sun, Yong; Zhu, Jigui; Tombari, Federico

doi:10.1109/lra.2021.3101523

Cited by 11 publications

(6 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides, Root Mean Square

{(RMS)}

is also a convictive indicator to evaluate the sphere fitting quality, reflecting the deviation degree of the distance from the spherical space point cloud to the center of the fitting sphere relative to the fitting radius. As shown in Table 6, our algorithm achieves the best results 0.52057 on the RMS index with more competitive computational efficiency than CFNet [22] and better accuracy than SRHNet [24], which indicates that our algorithm has more stable and robust 3D reconstruction effects in the ill‐posed realistic scenes.…”

Section: Applicationmentioning

confidence: 93%

“…EDNet [23] designs a combined volume and proposes attention-based spatial residual module for aggregating the combined volume. SRHNet [24] converts the cost volume into costs maps, and aggregates it with an hour-glass network consisting of SRH modules and GRUs. However, due to the local ambiguity of the depth mutation regions near the object edges, the above methods will lose boundaries details.…”

Section: End-to-end Stereo Matching Based On Cnnsmentioning

confidence: 99%

“…We deploy four competitive stereo matching methods in the system, including our GAMNet, GwcNet (2019) [7], SRHNet (2021) [24], PSMNet (2018) [6], and visualize the central point of each marker. As shown in Figure 17, the dispersion coefficient σ of GAMNet is 0.351%, surpassing GwcNet by 0.037% and SRHNet by 0.072%.…”

Section: Applicationmentioning

confidence: 99%

See 2 more Smart Citations

GAMNet: Global attention via multi‐scale context for depth estimation algorithm and application

Yang,

Lei,

Sang

2023

IET Image Processing

View full text Add to dashboard Cite

Deep neural networks significantly enhance the accuracy of the stereo‐based disparity estimation. Some current methods suffer from inefficient use of the global context information, which will lead to the loss of structural details in ill‐posed areas. To this end, a novel stereo network GAMNet is designed, composed of three core components (GDA, MPF, DCA) for estimating the depth prediction in challenging real‐world environments. First, a lightweight attention module is presented, integrating the global semantic cues for every feature position across the channel and spatial dimensions. Next, the MPF module is constructed to fuse the diverse semantic and contextual information from different levels of the feature pyramid. Finally, cost volume is aggregated with a stacked encoder‐decoder composed of the DCA module and 3D convolutions, filtering the transmission of matching clues and capturing the rich global contexts. Substantial experiments conducted on KITTI 2012, KITTI 2015, SceneFlow, and Middlebury‐v3 datasets manifest that GAMNet surpasses preceding methods with contour‐preserving disparity predictions. In addition, the first 3D scene reconstruction linear evaluation strategy on spatial grasping points for the end‐to‐end stereo networks in an unsupervised mode is proposed, and it is deployed on the designed robot vision‐guided system. In application experiments, the method can produce densely high‐precision 3D reconstructions to implement the grasping task in complex real‐world scenes, and achieve excellent robust performance with competitive inference efficiency.

show abstract

“…Besides, Root Mean Square

{(RMS)}

Section: Applicationmentioning

confidence: 93%

Section: End-to-end Stereo Matching Based On Cnnsmentioning

confidence: 99%

See 1 more Smart Citation

GAMNet: Global attention via multi‐scale context for depth estimation algorithm and application

Yang,

Lei,

Sang

2023

IET Image Processing

View full text Add to dashboard Cite

show abstract

“…OpenColorIO is an open source library available online [5]. The library requires configuration files encoding the different color transformations.…”

Section: Impact Of the Dataset Size On Evaluating Generalization Abilitymentioning

confidence: 99%

Active-Passive SimStereo -- Benchmarking the Cross-Generalization Capabilities of Deep Learning-based Stereo Methods

Jospin¹,

Antony²,

Liu³

et al. 2022

Preprint

View full text Add to dashboard Cite

In stereo vision, self-similar or bland regions can make it difficult to match patches between two images. Active stereo-based methods mitigate this problem by projecting a pseudo-random pattern on the scene so that each patch of an image pair can be identified without ambiguity. However, the projected pattern significantly alters the appearance of the image. If this pattern acts as a form of adversarial noise, it could negatively impact the performance of deep learning-based methods, which are now the de-facto standard for dense stereo vision. In this paper, we propose the Active-Passive SimStereo dataset and a corresponding benchmark to evaluate the performance gap between passive and active stereo images for stereo matching algorithms. Using the proposed benchmark and an additional ablation study, we show that the feature extraction and matching modules of a selection of twenty selected deep learning-based stereo matching methods generalize to active stereo without a problem. However, the disparity refinement modules of three of the twenty architectures (ACVNet, CascadeStereo, and StereoNet) are negatively affected by the active stereo patterns due to their reliance on the appearance of the input images.

show abstract

“…By promoting learning across different viewpoints, the method improves the stereo-matching performance. SRHNet [ 13 ] tackled the challenge of dealing with the 4D cubic cost volume used by 3D convolutional filters. It decouples the cost volume into sequential cost maps along the disparity direction, employing a recurrent cost aggregation strategy to handle it effectively.…”

Section: Introductionmentioning

confidence: 99%

Accuracy and efficiency stereo matching network with adaptive feature modulation

Lin,

Zhuo,

2024

PLoS ONE

View full text Add to dashboard Cite

Feature enhancement plays a crucial role in improving the quality and discriminative power of features used in matching tasks. By enhancing the informative and invariant aspects of features, the matching process becomes more robust and reliable, enabling accurate predictions even in challenging scenarios, such as occlusion and reflection in stereo matching. In this paper, we propose an end-to-end dual-dimension feature modulation network called DFMNet to address the issue of mismatches in interference areas. DFMNet utilizes dual-dimension feature modulation (DFM) to capture spatial and channel information separately. This approach enables the adaptive combination of local features with more extensive contextual information, resulting in an enhanced feature representation that is more effective in dealing with challenging scenarios. Additionally, we introduce the concept of cost filter volume (CFV) by utilizing guide weights derived from group-wise correlation. CFV aids in filtering the concatenated volume adaptively, effectively discarding redundant information, and further improving matching accuracy. To enable real-time performance, we designed a fast version named Fast-GFM. Fast-GFM employs the global feature modulation (GFM) block to enhance the feature expression ability, improving the accuracy and stereo matching robustness. The accurate DFMNet and the real-time Fast-GFM achieve state-of-the-art performance across multiple benchmarks, including Scene Flow, KITTI, ETH3D, and Middlebury. These results demonstrate the effectiveness of our proposed methods in enhancing feature representation and significantly improving matching accuracy in various stereo matching scenarios.

show abstract

SRH-Net: Stacked Recurrent Hourglass Network for Stereo Matching

Cited by 11 publications

References 26 publications

GAMNet: Global attention via multi‐scale context for depth estimation algorithm and application

GAMNet: Global attention via multi‐scale context for depth estimation algorithm and application

Active-Passive SimStereo -- Benchmarking the Cross-Generalization Capabilities of Deep Learning-based Stereo Methods

Accuracy and efficiency stereo matching network with adaptive feature modulation

Contact Info

Product

Resources

About