Vanet: a View Attention Guided Network for 3d Reconstruction from Single and Multi-View Images

Tang, Jilin; Zou, Zhengxia

doi:10.1109/icme51207.2021.9428171

Cited by 12 publications

(13 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The loss function consists of the absolute squared difference between the initial depth map and the GT (Ground Truth) depth map and the absolute squared difference between the refined depth map and the GT depth map, and the expression of the loss function is Eq. (8).…”

Section: Loss Functionmentioning

confidence: 99%

3D multi views reconstruction of flame surface based on deep learning

Song,

Liu,

Jiang

et al. 2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

A deep learning network based on MVSNet multi-view 3D reconstruction network (IM-MVSNet) is proposed in this paper for the reconstruction of the surface of laminar flame, which could suppress the influence of background noise on the reconstruction data when reconstructing the surface of laminar flame and improve the reconstruction accuracy of the flame surface at the same time. The network obtains high-quality segmented images by image segmentation of the reference frames and neighboring frames of the input sampled images to remove the background noise during the time of sampling, then reconstructs the multi-view images in 3D to build a 3D point cloud of the laminar flame surface, finally obtains the reconstructed laminar flame surface. The comparison of computational results using different reconstruction models shows that the 3D reconstruction network proposed in this paper could effectively reduce the point cloud noise of the reconstructed flame surface, improve the reconstruction accuracy of the flame surface, and provide a novel technical means for combustion research.

show abstract

Section: Loss Functionmentioning

confidence: 99%

3D multi views reconstruction of flame surface based on deep learning

Song,

Liu,

Jiang

et al. 2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

show abstract

“…Wen Chao et al established a multi-view deformation network based on Pixel2mesh network and realized the 3D reconstruction of neural network for graph convolutional [19]. Yuan et al added the auxiliary view features with weights to the main view features, deformed the preliminary reconstructed 3D model using the graph convolutional neural network, and gradually refined the grid model [20] combined with multiple image features. However, the multi-view 3D reconstruction method based entirely on deep learning performs poorly in reconstructing the model details.…”

Section: Introductionmentioning

confidence: 99%

Research on 3D reconstruction algorithms for small scenes with weak texture

Nai

2024

International Conference on Algorithm, Imaging Processing, and Machine Vision (AIPMV 2023)

View full text Add to dashboard Cite

3D reconstruction is 3 d reconstruction in aerial reconstruction, industrial measurement, medical image reconstruction, cultural relics preservation and restoration, virtual reality and other fields. However, the traditional 3 D reconstruction algorithm will have a poor reconstruction effect due to the transient foreground influence and the limitation of feature point identification in the scene. To solve the above problems, this paper uses LoFT R-SIFT algorithm to extract the feature points in weak texture area, increase the number of feature points matching in weak texture area, then introduces ExtremeC3Net algorithm to eliminate the feature points on the dynamic portrait in the scene; Finally, DPT improves the MVS algorithm to make deep compensation. The experimental results prove that the feature point matching accuracy of the algorithm is improved by 55%, which can better capture the details of the scene.

show abstract

“…Existing research on single-view reconstruction employs various 3D shape representations, including implicit functions [4][5][6][7], voxels [8][9][10][11], point clouds [12][13][14], meshes [15][16][17][18][19][20][21][22][23], and others. Voxel-based methods are challenging to reconstruct high-precision shapes due to the large amount of memory and computational time required to apply 3D CNN.…”

Section: Introductionmentioning

confidence: 99%

FAmesh: Generating Frequency Adaptive Meshes from Single Images under 2D Hole Constraints

Wen

2023

Applied Sciences

View full text Add to dashboard Cite

Reconstructing 3D models from a single image has numerous applications in fields such as VR/AR, medical imaging, and gaming. However, most mesh-based methods are limited by the use of 0-genus initial templates, which makes it difficult to reconstruct 3D meshes with complex topologies. Additionally, existing methods often prioritize reconstructing the overall shape and neglect to study local meshes with varying curvatures, resulting in a lack of correct and detailed local features in the generated meshes. This paper proposes a 3D reconstruction framework that transitions from global to local and incorporates MLP and GCN. The framework introduces a mesh pruning strategy under a 2D hole constraint to restore the correct mesh topology. Moreover, the framework fine-tunes local details by separately learning corresponding mapping functions on high-frequency and low-frequency local extended patches. The experiment with the proposed network on the ShapeNet dataset shows that the network has a CD value of 1.763 and an F-score of 85.40. The results from extensive experiments demonstrate that our proposed method outperforms existing methods in topology correction and local detail reconstruction.

show abstract

Vanet: a View Attention Guided Network for 3d Reconstruction from Single and Multi-View Images

Cited by 12 publications

References 13 publications

3D multi views reconstruction of flame surface based on deep learning

3D multi views reconstruction of flame surface based on deep learning

Research on 3D reconstruction algorithms for small scenes with weak texture

FAmesh: Generating Frequency Adaptive Meshes from Single Images under 2D Hole Constraints

Contact Info

Product

Resources

About