KNOWLEDGE ASSET
Automated Train-Rivet Crack Detection Using Fluorescent Magnetic Particle Inspection and Instance Segmentation
Original title：sr 2024 fluorescent mpi rivets
Professional source metadata, version and usage boundary.
Curated summary
This open-access paper automates crack detection in fluorescent magnetic-particle images of train rivets using instance segmentation, dense-crack post-processing, channel-spatial attention, and multi-task feature learning. It also develops a non-contact composite magnetization device for different rivet shapes. The reported recall and mAP at 0.5 are 86.4 and 90.3 percent. The approach may inform spring-surface inspection, but its dataset and workpiece differ and require independent validation.
Original language：en
Bibliographic metadata
| File | sr-2024-fluorescent-mpi-rivets.pdf |
|---|---|
| Type / category | reference_pdf / references |
| Section | Installation & Use |
| Organization | — |
| Authors | — |
| Language | en |
| Rights / access | open_license / public |
Version
Version not identified · 1/1/2024 · active · current
Professional identifiers
Identifier pending review.