연구 자료(Research Data)
1) Repository for segmented point clouds of fitting parts in the process plants for deep learning application
- Number of point clouds: 4,647 segmented point clouds for 17 fitting part types:
699 segmented point clouds for valve-type fittings, 1,620 for the flange series, 15 for the cross series,
507 for the tee series, 18 for olet (branch) types, 75 for strainer types, 300 for the reducer series,
1,341 for the elbow series, and 72 for pipe types
- File type: XYZ format
- Number of points in point clouds: from around 360 points to around 1.4 million points for each segmented point cloud
- File size: from 12 kilobytes to 42 megabytes for each segmented point cloud
For more information on the repository, please refer to the following paper.
- Changmo Yeo, Seyoon Kim, Hyungki Kim, Siro Kim, and Duhwan Mun (Corresponding Author), “Deep learning applications in an industrial process plant: Repository of segmented point clouds for pipework components”, JMST Advances, Vol. 2, No. 1, pp. 15-24, 2020.03.01.
If you want to download the set of segmented point clouds of fitting parts, click here!
2) Dataset for 3D CAD models containing features in OBJ format
- Number of 3D CAD models: 119,320 3D CAD models in two versions:
version 1) 101,000 3D CAD models, two types of features(hole, pocket)
version 2) 18,320 3D CAD models, four types of features(chamfer, fillet, hole, pocket)
- File type: OBJ format
- File size: from 4 kilobytes to 251 kilobytes for each 3D CAD model
For more information on the repository, please refer to the following papers.
- Hyunoh Lee, Jinwon Lee, Hyungki Kim, Duhwan Mun, “Deep learning-based reconstruction of 3D CAD models containing machining features for mechanical parts”, Journal of Computational Design and Engineering, In Review, 2021.
If you want to download the set of 3D CAD models containing machining features, click here!
3) Dataset of feature descriptors used for machining feature recognition
- Number of feature descriptors: 2,236 feature descriptors for 17 types
406 descriptors for the hole series, 722 for the slot series, 380 for the pocket series, 356 for the island series,
182 for the fillet series, 46 for the chamfer series, and 174 for the non-feature types
- File type: TXT format
- File size: 122 bytes for each feature descriptor
For more information on the dataset, please refer to the following papers.
- Changmo Yeo, Byung Chul Kim, Sanguk Cheon, Jinwon Lee, and Duhwan Mun, "Machining feature recognition based on deep neural networks to support tight integration with 3D CAD systems", Scientific Reports, In Review, 2021.
If you want to download the dataset based on feature descriptors, click here!