Review of multi-view 3D object recognition methods based on deep learning

Qi, Shaohua; Ning, Xin; Yang, Guowei; Zhang, Liping; Peng, Long; Cai, Weiwei; Li, Weijun

doi:10.1016/j.displa.2021.102053

Cited by 140 publications

(38 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An optimal alignment procedure, akin to rigidly rotating the positions of the virtual cameras, is performed to compare two shapes with minimum distortion. Advancements in view-based shape recognition include hypergraph representations of the captured 2D images [38], and various learning-based processes, summarized by [29].…”

Section: Shape Similaritymentioning

confidence: 99%

Spherical harmonic shape descriptors of nodal force demands for quantifying spatial truss connection complexity

Lee

Danhaive

Mueller

2022

Archit. Struct. Constr.

View full text Add to dashboard Cite

The connections of a spatial truss structure play a critical role in the safe and efficient transfer of axial forces between members. For discrete connections, they can also improve construction efficiency by acting as registration devices that lock members in precise orientations. As more geometrically complex spatial trusses are enabled by computational workflows and the demand for material-efficient spanning systems, there is a need to understand the effects of global form on the demands at the connections. For large-scale structures with irregular geometry, customizing individual nodes to meet exact member orientations and force demands may be infeasible; conversely, standardizing all connections results in oversized nodes and a compromise in registration potential. We propose a method for quantifying the complexity of spatial truss designs by the variation in nodal force demands. By representing nodal forces as a geometric object, we leverage the spherical harmonic shape descriptor, developed for applications in computational geometry, to characterize each node by a rotation and translation-invariant fixed-length vector. We define a complexity score for spatial truss design by the variance in the positions of the feature vectors in higher-dimensional space, providing an additional performance metric during early stage design exploration. We then develop a pathway towards reducing complexity by clustering nodes with respect to their feature vectors to reduce the number of unique connectors for design while minimizing the effects of mass standardization.

show abstract

Section: Shape Similaritymentioning

confidence: 99%

Spherical harmonic shape descriptors of nodal force demands for quantifying spatial truss connection complexity

Lee

Danhaive

Mueller

2022

Archit. Struct. Constr.

View full text Add to dashboard Cite

show abstract

“…Fifthly, according to the rules of evidential reasoning, the prediction results of the base classifier are synthesized to get the final prediction results. classification algorithm, and its characteristic is that it can solve high-dimensional and non-linear classification problems by using the principle of structural risk minimization and can efficiently classify data samples with few samples and high feature dimensions [16,17]. erefore, SVM is chosen as the base classifier of the financial risk prediction model.…”

Section: Construction Of Financial Risk Prediction Modelmentioning

confidence: 99%

Research on Financial Risk Prediction Based on Improved Random Subspace

2022

Scientific Programming

View full text Add to dashboard Cite

In order to provide timely and effective information and decision support for financial market entities, combined with random subspace and weight fused Lasso, this paper constructs a financial risk prediction model based on the improved random subspace method. Firstly, the basic principles of random subspace and SVM algorithm are introduced. Then, WFL and Al methods are introduced to improve random subspace, so as to reduce the dimension of multisource heterogeneous data and realize the adaptive fusion of features. Then, a financial risk prediction model based on weighted fusion adaptive random subspace is constructed, in which SVM is used as the basic classifier and the output strategy of result integration is introduced. Finally, based on the data of some listed companies, the improved random subspace method is compared with other methods. The results show that the improved random subspace method has a higher prediction value, which indicates that the method is reasonable and effective in financial risk prediction. In the improved random subspace method, combined feature F1 + F2 + F3 is better than other methods in T − 3, T − 4, and T − 5, and the prediction value is more than 95%, which fully demonstrates the rationality of the improved random subspace method in financial risk prediction. The area under the ROC curve (AUC) predicted by weight fused adaptive integration-based random subspace (FAIB_RS) method is about 95% in T − 3, 93% in T − 4, and 95.5% in T − 5, which is obviously higher than that of the other eight methods.

show abstract

“…Deep learning approaches are categorized depending on the representation the object data as: (1) Voxelbased methods, e.g., VoxNet [31], ShapeNets [16], VoxelNet [32], or (2) Point-set-based methods, e.g., PointNet [33], FoldingNet [34] and, finally, (3) View-based methods, e.g., RotationNet [35], HMVCM [36] and Complex-YOLO [37]. Voxel-based and Point-setbased methods, also referred to as model-based methods [38], use a 3D volumetric CNN, and therefore, they exploit the 3D geometry of the objects. However, volumetric CNNs have a large and complex CNN architecture and require high computational and memory resources, and these are therefore not suitable for real-time applications.…”

Section: Related Workmentioning

confidence: 99%

“…View-based methods, instead, transform the 3D objects into a series of 2D images from different viewpoints. As a consequence, they use 2D CNN methods and do not fully exploit the 3D data geometry of the objects, although they achieve good recognition performances, especially in the case of occlusions [38]. An alternative method for object recognition by a depth camera is to include the depth channel along RGB channels (RGB-D) in combination with a 2D CNN [39] and recursive neural networks (RNNs) [40] or encode the depth channel in jet color maps and the surface of normals [41].…”

Section: Related Workmentioning

confidence: 99%

Evaluation of 2D-/3D-Feet-Detection Methods for Semi-Autonomous Powered Wheelchair Navigation

et al. 2021

View full text Add to dashboard Cite

Powered wheelchairs have enhanced the mobility and quality of life of people with special needs. The next step in the development of powered wheelchairs is to incorporate sensors and electronic systems for new control applications and capabilities to improve their usability and the safety of their operation, such as obstacle avoidance or autonomous driving. However, autonomous powered wheelchairs require safe navigation in different environments and scenarios, making their development complex. In our research, we propose, instead, to develop contactless control for powered wheelchairs where the position of the caregiver is used as a control reference. Hence, we used a depth camera to recognize the caregiver and measure at the same time their relative distance from the powered wheelchair. In this paper, we compared two different approaches for real-time object recognition using a 3DHOG hand-crafted object descriptor based on a 3D extension of the histogram of oriented gradients (HOG) and a convolutional neural network based on YOLOv4-Tiny. To evaluate both approaches, we constructed Miun-Feet—a custom dataset of images of labeled caregiver’s feet in different scenarios, with backgrounds, objects, and lighting conditions. The experimental results showed that the YOLOv4-Tiny approach outperformed 3DHOG in all the analyzed cases. In addition, the results showed that the recognition accuracy was not improved using the depth channel, enabling the use of a monocular RGB camera only instead of a depth camera and reducing the computational cost and heat dissipation limitations. Hence, the paper proposes an additional method to compute the caregiver’s distance and angle from the Powered Wheelchair (PW) using only the RGB data. This work shows that it is feasible to use the location of the caregiver’s feet as a control signal for the control of a powered wheelchair and that it is possible to use a monocular RGB camera to compute their relative positions.

show abstract

Review of multi-view 3D object recognition methods based on deep learning

Cited by 140 publications

References 45 publications

Spherical harmonic shape descriptors of nodal force demands for quantifying spatial truss connection complexity

Spherical harmonic shape descriptors of nodal force demands for quantifying spatial truss connection complexity

Research on Financial Risk Prediction Based on Improved Random Subspace

Evaluation of 2D-/3D-Feet-Detection Methods for Semi-Autonomous Powered Wheelchair Navigation

Contact Info

Product

Resources

About