A naukri.com initiative
Bioengineer
6d
343
Image Credit: Bioengineer
Unsupervised Porosity Segmentation in Laser Powder Fusion
- A groundbreaking study introduces an unsupervised machine learning methodology for porosity segmentation in laser powder bed fusion (LPBF) parts.
- The approach utilizes the 'Segment Anything' model to automate porosity identification without extensive labeled datasets.
- LPBF faces porosity issues due to factors like improper parameters, contamination, or gas entrapment, prompting the need for efficient defect detection.
- Traditional inspection methods are costly and time-consuming, while manual defect labeling hinders scalability.
- The unsupervised approach detects porosity based on image contrasts and textures, eliminating the need for annotated datasets.
- Operators can interactively guide the segmentation process with minimal input, enhancing adaptability and defect detection efficiency.
- Advanced technical innovations empower the model to distinguish between true porosity and artifacts, improving defect identification accuracy.
- Accurate porosity mapping aids in predictive maintenance, process optimization, and enhancing the quality and reliability of manufactured parts.
- The model's promptability, efficiency, and adaptability make it a valuable tool for various LPBF platforms and materials, enhancing defect surveillance.
- The methodology demonstrated superior segmentation accuracy over supervised models, showcasing robustness against noise and scan resolution variations.
Read Full Article
20 Likes
For uninterrupted reading, download the app