Automating Porosity Detection with Deep Learning
summary
Claims that CNN segmentation models plus EfficientNet transfer learning convert X‑ray CT stacks into pixelwise porosity maps, spotting pores missed by human inspection and enabling automatic accept/reject thresholds. Workflow lists concrete steps: 3D→2D slice conversion, noise filtering, intensity normalization, augmentation, and training with Adam/AdamX optimizers on V‑Net/segmentation backbones. Data comes from lab‑scanned parts, simulated CT and hybrid real+synthetic CT data; limitations are costly compute and poor cross‑material generalization (plastic models fail on metal).
AI-written summary · about
chapters (11)
- 0:00Additive manufacturing basics
- 3:17Porosity detection importance
- 6:27Traditional inspection limits
- 8:39Deep learning automation
- 10:49Data acquisition challenges
- 13:03CT scan preprocessing
- 15:58Data annotation & augmentation
- 17:25Classification vs segmentation
- 19:13Transfer learning benefits
- 20:37Model accuracy evaluation
- 21:42Limitations & future work
related talks
23mPermeability Prediction of Porous Media using CNN with Physical Properties
08mDeep Transfer Learning for Computer Vision
26mTraining a TensorFlow model to detect lung nodules on CT scans
43mBuilding AI That Learns and Adapts: A Case Study in MRI Diagnostics
55mDeep Learning in Medical Image Diagnostics
11mInterview
27mU-Net is all you need: applying the same AI approach for many different business domains
49mBuilding AI That Learns and Adapts: A Case Study in MRI Diagnostics