FabWave CAD Repository Categorized Part Classes - CAD 16 through 24 Classes
Abstract
This is Part 2 of the 3D Engineering CAD data in Part 2 - Classes 16 through 24. The classes are standardized components - Pipe fittings, Pipe joints, Pipes, Rollers, Rotary Shaft, Shaft Collar, Slotted Flat Head Screws, Socket Head Screws, Washers
File (1)
Content uploaded by Binil Starly
Author content
... Nevertheless, a robust center calculation is crucial for the method. Tampering with the model's center based on the Fabwave dataset [23] results in an average similarity of 92% for a 1% randomized scalar applied to the model's center of mass. However, shifting the sphere center against our preprocessing requires significant tampering with the mesh, which underpins this approach's robustness. ...
... The performance of the OH and the related shape discriminators were tested against modifications of rotation, uniform scaling, deform scaling, and collisions with the results in Table 1. The Fabwave dataset contains industrial components such as screws and nuts with 68 models from the uncategorized Fabwave [23]. Furthermore, the OH was successfully tested for collisions within a larger dataset of 1000 models from the ShapeNet dataset with no occurring hash collisions Table 3 All methods performed notably well against rotations and uniform scaling. ...
... Repeating the rotation test with the OpenShape framework, we observe a high sensitivity to small rotations. For example, we examined the classification process with the Fabwave dataset [23]. Rotating the models 90 • around the x-axis causes the resulting top five categories to change completely. ...
Deep neural networks have shown promising success towards the classification and retrieval tasks for images and text data. While there have been several implementations of deep networks in the area of computer graphics, these algorithms do not translate easily across different datasets, especially for shapes used in product design and manufacturing domain. Unlike datasets used in the 3D shape classification and retrieval in the computer graphics domain, engineering level description of 3D models do not yield themselves to neat distinct classes. The current study looks at an improved form of the 3D shape deep learning algorithm for classification and retrieval through the use of techniques such as relaxed classification, use of prime angled camera angles for capturing feature detail and transfer learning for reducing the amount of data and processing time needed to train shape recognition algorithms. The proposed algorithm (MVCNN++) builds on top of multi-view convolutional neural network (MVCNN) algorithm, improving its efficacy for manufacturing part classification by enabling use of part metadata, yielding an improvement of almost 6% over the original version. With the explosive growth of 3D product models available in publicly available repositories, search and discovery of relevant models is critical to democratizing access to design models.
ResearchGate has not been able to resolve any references for this publication.