Summary of Attenuation-adjusted Deep Learning Of Pore Defects in 2d Radiographs Of Additive Manufacturing Powders, by Andreas Bjerregaard and David Schumacher and Jon Sporring
Attenuation-adjusted deep learning of pore defects in 2D radiographs of additive manufacturing powders
by Andreas Bjerregaard, David Schumacher, Jon Sporring
First submitted to arxiv on: 5 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG); Image and Video Processing (eess.IV)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents a novel approach to analyze the porosity of metal feedstock powders used in additive manufacturing. The proposed method simulates efficient pore segmentation by labeling pore pixels on single 2D radiographs of powders. This is achieved through a combination of X-ray attenuation modeling and a variant of the UNet architecture, which boosts F1-score by 11.4% compared to the baseline UNet. The model leverages pretraining on synthetic data, tight particle cutouts, and subtracting an ideal particle without pores generated from a distance map inspired by Lambert-Beers law. The paper evaluates four image processing methods, with the fastest method segmenting a particle in 0.014s with F1-score 0.78, while the most accurate method takes 0.291s with F1-score 0.87. These strategies can be scaled for high-throughput porosity analysis of metal feedstock powders. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about finding a way to quickly and accurately analyze the tiny holes (pores) in metal powder used to make objects using a special process called additive manufacturing. Right now, it takes a long time to look at these pores, but this research tries to find a faster way. They did this by creating a computer program that can look at a 2D picture of the powder and tell where the pores are. This program is really good at finding the right spots – better than what’s being used now! The researchers tried different ways to make their program work, and they found one that was both fast and accurate. This could be useful in making more objects using this special process. |
Keywords
» Artificial intelligence » F1 score » Pretraining » Synthetic data » Unet
