SSRNet: Scalable 3D Surface Reconstruction Network

Mi, Zhenxing; Luo, Yiming; Tao, Wenbing

doi:10.1109/cvpr42600.2020.00105

Cited by 61 publications

(32 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Despite voxel-based approaches being easy to represent and train due to low mathematical complexity of loss function, they suffer from a major flaw: the exponential memory requirements in order to train high granularity model, which would be required in order to reconstruct complex models containing a lot of details. While there have been attempts to solve this issue of ever-increasing memory requirements by using more compact data representation styles, such as octrees [ 30 , 31 ], thus greatly reducing the amount of required data to represent the same model, these still suffer from overheads.…”

Section: Related Workmentioning

confidence: 99%

Auto-Refining Reconstruction Algorithm for Recreation of Limited Angle Humanoid Depth Data

Kulikajevas

Maskeliūnas

Damaševičius

et al. 2021

Sensors

View full text Add to dashboard Cite

With the majority of research, in relation to 3D object reconstruction, focusing on single static synthetic object reconstruction, there is a need for a method capable of reconstructing morphing objects in dynamic scenes without external influence. However, such research requires a time-consuming creation of real world object ground truths. To solve this, we propose a novel three-staged deep adversarial neural network architecture capable of denoising and refining real-world depth sensor input for full human body posture reconstruction. The proposed network has achieved Earth Mover and Chamfer distances of 0.059 and 0.079 on synthetic datasets, respectively, which indicates on-par experimental results with other approaches, in addition to the ability of reconstructing from maskless real world depth frames. Additional visual inspection to the reconstructed pointclouds has shown that the suggested approach manages to deal with the majority of the real world depth sensor noise, with the exception of large deformities to the depth field.

show abstract

Section: Related Workmentioning

confidence: 99%

Auto-Refining Reconstruction Algorithm for Recreation of Limited Angle Humanoid Depth Data

Kulikajevas

Maskeliūnas

Damaševičius

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Wang et al, 2018) For 3D reconstruction, some methods generate the surface from a point cloud constructed from RGB-D images. This reconstruction can be without texture mapping (F. Wang & Hauser, 2019), (Sheng et al, 2018), (Tzionas & Gall, 2015), (Gao et al, 2019), (K. Wang et al, 2014), (Zhong et al, 2019), (Mi et al, 2020), and(Kazhdan et al, 2013) or with texture mapping (Vrubel et al, 2009) and (Tucci et al, 2012). Some methods use a priori templates (Hao et al, 2019).…”

Section: Generation Of Surface Modelsmentioning

confidence: 99%

3D Reconstruction and Texture Refinement with Image Enhancement for Improved Visual Quality

Corona-Patricio¹,

Reta²

2021

EJT

View full text Add to dashboard Cite

This paper proposes a method to generate surface and texture models from rigid objects captured with an RGB-D camera. The method integrates five stages: 1. Point cloud generation from RGB-D images; 2. Surface model generation; 3. Surface model refinement; 4. Texture generation and mapping; 5. Texture enhancement. The use of image processing algorithms for texture enhancement and the refinement of the surface models enables the improvement of the appearance of reconstructed models. The performed experimentation shows the results of the proposed method for five small textured objects. The appearance of reconstructed models was evaluated using a visual quality index; a sharper texture helps to improve such index.

show abstract

“…There have been extensive studies on single image based 3-D object reconstruction with various ways for 3D shape representation, including voxels, octree [30,35,40], deep implicit function, mesh and point cloud [6,17,29,23]. Methods based on different representations all have their cons and pros.…”

Section: Single Image Based Object Reconstructionmentioning

confidence: 99%

Object Wake-up: 3D Object Rigging from a Single Image

Yang¹,

Zuo²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Given a picture of a chair, could we extract the 3-D shape of the chair, animate its plausible articulations and motions, and render in-situ in its original image space? The above question prompts us to devise an automated approach to extract and manipulate articulated objects in single images. Comparing with previous efforts on object manipulation, our work goes beyond 2-D manipulation and focuses on articulable objects, thus introduces greater flexibility for possible object deformations. The pipeline of our approach starts by reconstructing and refining a 3-D mesh representation of the object of interest from an input image; its control joints are predicted by exploiting the semantic part segmentation information; the obtained object 3-D mesh is then rigged & animated by non-rigid deformation, and rendered to perform in-situ motions in its original image space. Quantitative evaluations are carried out on 3-D reconstruction from single images, an established task that is related to our pipeline, where our results surpass those of the SOTAs by a noticeable margin. Extensive visual results also demonstrate the applicability of our approach.

show abstract

SSRNet: Scalable 3D Surface Reconstruction Network

Cited by 61 publications

References 20 publications

Auto-Refining Reconstruction Algorithm for Recreation of Limited Angle Humanoid Depth Data

Auto-Refining Reconstruction Algorithm for Recreation of Limited Angle Humanoid Depth Data

3D Reconstruction and Texture Refinement with Image Enhancement for Improved Visual Quality

Object Wake-up: 3D Object Rigging from a Single Image

Contact Info

Product

Resources

About