Graph-augmented Segmentation for Enhanced In-Situ Geometry Reconstruction of Complex Shapes in Laser Powder Bed Fusion
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: AI Radio. Generated commentary on the latest Artificial Intelligence papers.
Tom: We're looking at the paper: "Graph-augmented Segmentation for Enhanced In-Situ Geometry Reconstruction of Complex Shapes in Laser Powder Bed Fusion".
Title and authors: Tom: So we're looking at this paper today, "Graph-augmented Segmentation for Enhanced In-Situ Geometry Reconstruction of Complex Shapes in Laser Powder Bed Fusion." It’s tackling that big problem of reliably inspecting things made by laser powder bed fusion.
Jane: That's right. The paper is specifically focused on how to do image segmentation of powder bed images in laser powder bed fusion, which is super hard because the lighting and the intensity patterns change from one layer to the next.
Lu: Basically, they’re addressing those limitations by proposing a graph-augmented segmentation approach. The core idea is shifting how they capture information, moving away from just looking at individual pixels to modeling dependencies across the whole image globally.
Meng: So what does that actually mean for us on the shop floor? Does it mean we can finally get reliable in situ inspection without needing expensive destructive testing every single time?
Tom: Exactly. The paper highlights that current methods struggle with illumination conditions and layer-to-layer variability, which is totally normal in L-PBF processes, so this graph augmentation is intended to make the segmentation more robust against that noise.
Jane: It sounds like they are trying to decouple the structural consistency of the part from those local intensity fluctuations that plague standard segmentation methods. That’s a really clever way to look at it.
Lu: They formulate a graph where nodes are pixels and edges connect neighboring pixels based on their distance, using a k-nearest neighbor formulation, so each node features in layer t are updated by aggregating and weighting info from neighbors in layer t minus one. This creates this relational structure the network learns from.
Meng: Okay, so they’re building this graph structure inside a U-Net bottleneck. That sounds complex to implement at scale for production systems, but if it works better than the current U-Net baselines, it might be worth the integration effort.
Tom: The authors show that this UNet-GNN architecture actually has higher accuracy than both the standard U-Net and active contour methods, even when things aren't perfectly matched between training and testing data.
Jane: That's a big deal because if it performs better across different build locations or scan directions, it means the reconstruction of the geometry is more consistent regardless of where you are in the machine.
Lu: They also found that this UNet-GNN is less sensitive to geometry variations from layer to layer and scan direction variations compared to those other two methods, which are intrinsic problems in L-PBF processes.
Meng: That robustness against those process variations is what I’m really interested in for deployment because if the model doesn't constantly need retraining when we change the build parameters slightly, that saves a lot of engineering time.
Tom: And they also tested it with a gamma perturbation, which messes with global contrast and brightness properties, and while both models degraded, the UNet-GNN remained systematically more accurate than the competitors.
Title and authors: Jane: That suggests it’s capturing something about the overall structure better than just local pixel intensity changes when things get visually messy.
Lu: When Gaussian noise is added to the image, a standard U-Net shows an abrupt drop in segmentation accuracy, but the UNet-GNN yields higher and more stable performance under that kind of noise.
Meng: Stability is key for production monitoring; we don't want our quality control flagging an error because of a random speck of dust on the sensor.
Tom: And let's talk about speed here, because accuracy doesn't matter if it takes too long to run during the inspection process itself. Active contour methods are slow, taking about sixteen to seventeen seconds per image.
Jane: So this new approach is significantly faster for inference, right?
Lu: Yeah, the UNet-GNN provides a much faster inference time compared to the U-Net and active contour methods; it's about zero point four seconds per image on average.
Meng: Zero point four seconds is pretty fast for a complex model. That’s crucial if we need real-time monitoring during the build process itself, not just after.
Tom: So to wrap up this part, the UNet-GNN successfully integrates graph learning into the U-Net bottleneck to explicitly model spatial and relational dependencies in powder bed images.
Jane: And the results show it consistently outperforms both active contour methods and standard U-Net architectures across all tested scenarios when validated against ground truth data from X-ray CT of trabecular lattice structures.
Lu: The key finding is that this robustness represents a necessary feature for enabling in situ inspection as a reliable tool to support process qualification and product verification in industrially compliant implementations, according to the paper.
Meng: From my side, if we can get this level of reliability across different build locations and layers, it drastically cuts down on the need for costly post-process checks, which is a huge operational win.
Tom: It sounds like they’ve given us a strong candidate for making in situ geometry reconstruction much more reliable than what we have currently available.
Jane: So as we move forward, they're looking ahead to exploring how this method performs when real defects are present, which is the next logical step for applying it in a manufacturing setting.
Lu: That exploration into real defects is the natural progression because theoretical robustness is one thing, but handling actual material flaws requires testing against those specific edge cases.
Meng: I hope they don't just stop at simulation; they need to show us how this holds up when we actually introduce a tiny crack or a porosity in the printed part.
Tom: Well, for now, it looks like this graph-augmented segmentation for enhanced in situ geometry reconstruction of complex shapes in laser powder bed fusion is a solid step toward making AM qualification more efficient and trustworthy.
Jane: That's what we've covered today on this paper. We’ll be ready to dive into whatever is next on the arXiv list soon.
The paper's summary: Tom: So, to recap, this paper is about using graph learning inside a U-Net to make segmentation of powder bed images in laser powder bed fusion more reliable when things change layer by layer or with different lighting.
Jane: Right, and what they found is that this approach handles the noise much better than standard methods. They're essentially decoupling the actual shape of the part from all those random intensity fluctuations that happen as you print.
Lu: Exactly. The core idea is building a graph structure where pixels connect based on how close they are, and then updating their information layer by layer using that network structure instead of just looking at immediate neighbors. It lets the model see the whole context globally.
Meng: So what does this actually mean for me when I'm trying to deploy a monitoring system? Does it mean less time spent chasing false alarms because of a stray piece of dust on the sensor?
Tom: Yeah, it means higher accuracy even when the training and testing data are messy or inconsistent. They showed that this UNet-GNN consistently beats both the standard U-Net and active contour methods across all their tests.
Jane: That's huge because they tested it under really varied conditions, like switching between different build lattices for training versus testing. The accuracy gap with the standard U-Net gets much wider when things are not consistent.
Lu: They also looked at variations caused by the scan direction itself, and again, the UNet-GNN stayed more stable than those other two methods under those kinds of process changes. It shows it learns a more fundamental structure.
Meng: And on the speed side, they didn't sacrifice too much performance. The UNet-GNN is about zero point four seconds per image on average, which is way faster than the active contour method that takes sixteen to seventeen seconds. That computational efficiency makes it feasible for real-time monitoring during the print process.
Tom: And when they added Gaussian noise to the image—which mimics real sensor imperfections—the standard U-Net accuracy dropped sharply, but the UNet-GNN performance stayed high and stable. That stability against random noise is a big win for any production system.
Jane: It seems like they've built a model that doesn't just react to local pixel changes; it understands the relationship between different parts of the image to reconstruct the overall geometry accurately.
Lu: The implication here is that you can use this architecture to create in situ inspection tools that are much more robust against those inherent, layer-to-layer process variations common in additive manufacturing.
Tom: So, if we’re looking at qualifying parts or verifying product quality in these AM processes, this UNet-GNN gives us a much more reliable way to check the geometry without getting bogged down by image noise or build inconsistencies.
Jane: That's what they're aiming for. They proved that modeling spatial and relational dependencies makes the segmentation system far more dependable than previous methods.
Meng: I see how that translates to reducing the need for constant manual checks after every layer; if the model is this robust, it can give us much more consistent data throughout the whole build.
Tom: And they're not just proving it on perfect data; they validated it against ground truth from X-ray CT scans of trabecular lattice structures, which means the results hold up when you actually look at what's happening inside the part.
Jane: It’s a solid piece of work that moves us closer to having truly reliable, scalable in situ inspection methods for additive manufacturing. So, next time we talk about monitoring complex shapes in three dee printing, this UNet-GNN is definitely on our radar.
The paper's improvements: Tom: So, focusing on what they suggest as improvements, they are really pushing for that explicit modeling of spatial and relational dependencies within the bottleneck layer itself.
Jane: They’re saying the architecture needs to learn not just what a pixel looks like locally, but how it relates to its neighbors across the whole image structure.
Lu: Right, they want to move beyond just local appearance cues; they want the model to grasp higher-order connectivity patterns and structural constraints that define the object's shape. That graph formulation is key for that relational learning.
Meng: If we implement this, it means our segmentation isn't just a pixel-by-pixel classification anymore; it’s understanding the structure of the whole part simultaneously, which should lead to much more consistent results during monitoring.
Tom: And they showed that this structural context modeling directly leads to enhanced accuracy when facing those messy photometric variations we talked about earlier. It makes the segmentation robust against lighting changes and layer-to-layer intensity shifts.
Jane: So, the improvement isn't just better numbers; it’s a deeper understanding of *why* certain pixels are connected, which is what lets it handle those complex real-world manufacturing conditions.
Lu: The paper suggests that this explicit graph learning directly combats the sensitivity to geometry variations from one layer to the next and scan direction changes. It stabilizes the results across different acquisition conditions.
Meng: That would be huge for deployment because it means less of a calibration nightmare when we change our build parameters or move a machine around; the model adapts better inherently.
Tom: And they showed that when contrast properties change globally, like with a gamma perturbation, this architecture is systematically more accurate than the standard U-Net and active contour methods. It maintains its integrity under global image distortions.
Jane: That tells us the model isn't just brittle; it has a better mechanism for handling those broader environmental or process changes that affect the whole image at once.
Lu: The authors are basically arguing that integrating this graph representation into the U-Net bottleneck is a necessary step to ensure reliability in in situ inspection, proving its superior performance across all their tests.
Tom: The implication is clear: for any industrial application using L-PBF where geometry reconstruction needs to be dependable, this hybrid approach offers a much stronger foundation than just sticking with the standard U-Net.
Jane: It’s about moving from an image processing tool that reacts to local features to one that understands the underlying physical structure of the part.
Meng: From my side, if we can get this level of consistent performance in a production environment, it drastically cuts down on the need for costly post-process checks because the geometry reconstruction is already more reliable.
Lu: The paper does point out a limitation though; they haven't fully explored what happens when real physical defects—like cracks or porosity—are present in the part yet. That’s definitely the next area for development.
Tom: Right, that’s the caveat; theoretical robustness is one thing, but handling actual material flaws requires testing against those specific edge cases we need to address next.
Jane: So they've built a very strong foundation here, and now the next step is to see how this structure handles genuine imperfections in the printed metal itself.
Conclusion: Tom: So, to wrap up on this paper, "Graph-augmented Segmentation for Enhanced In-Situ Geometry Reconstruction of Complex Shapes in Laser Powder Bed Fusion," they've shown how graph learning inside a U-Net can significantly boost robustness against all those messy real-world imaging problems.
Jane: Right, the main implication is that we can build much more reliable in situ inspection tools for additive manufacturing because this method handles the layer-to-layer variability and lighting issues much better.
Lu: It’s a solid framework because it explicitly models the spatial relationships between image regions, which is what lets it achieve that higher segmentation accuracy even when data conditions aren't perfectly matched.
Meng: For production, this means we can trust the geometry reconstruction more during the build process itself, reducing errors from inconsistent acquisition and cutting down on post-process failures.
Lalam: I'm confident in this direction; the evidence shows it handles noise and contrast shifts much better than previous methods we’ve reviewed. It’s a very sure answer based on their rigorous testing against X-ray CT data.
Tom: Exactly, and the speed is good too; they keep inference time low at about zero point four seconds per image, which is important for real-time monitoring during a print run.
Jane: So, if you're driving or walking and you need to check a complex part made by laser powder bed fusion, this graph augmented approach gives us a much more dependable way to reconstruct the shape without getting fooled by image noise.
Lu: The main limitation they flag is that they haven't yet fully explored how this method performs when there are actual physical defects like cracks or porosity in the part itself.
Meng: That’s fair; it’s great for detecting geometric deviations, but testing against real material flaws is definitely the next challenge we need to tackle.
Tom: Well, that's a good point, and it leads us right into our next topic: exploring how this method handles genuine defects in the metal itself.
Stefano Raimondo, Matteo Bugatti, Marco Grasso
Politecnico di Milano
cs.CV
Submitted: 2026-04-27
Updated: 2026-11-18
Comments: Preprint
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Importance score: 91/100
The gist: Graph-augmented segmentation of complex shapes in laser powder bed fusion for enhanced in situ inspection is proposed to overcome limitations in existing methods concerning photometric variability…
Key concepts
- Graph Neural Network (GNN) Bottleneck
- A specific design feature added to the U-Net architecture where image regions are treated as a graph. Nodes represent pixel locations, and edges connect neighboring pixels based on their spatial proximity. This allows the network to learn how different parts of the powder bed relate to each other spatially, capturing complex structural dependencies beyond simple local pixel features.
- Photometric Variability
- The inconsistency in image appearance caused by industrial lighting conditions and layer-to-layer intensity patterns during L-PBF. This variability makes standard segmentation models struggle because the same physical structure looks different depending on when or where it was scanned, requiring a more resilient approach to maintain accurate reconstruction.
- UNet-GNN Architecture
- A hybrid deep learning model combining the U-Net's encoder-decoder structure with a GNN. The GNN component aggregates information from neighboring nodes in each layer, allowing the model to explicitly model spatial and relational dependencies within the powder bed geometry, which significantly boosts its ability to handle complex shapes reliably.
- In Situ Inspection
- The process of examining or monitoring the additive manufacturing process while it is actively occurring. This method uses image segmentation to reconstruct the actual 3D geometry of the printed part in real-time, enabling immediate detection of geometric deviations without needing post-process measurements.
Terminology
Summary
Graph-augmented segmentation of complex shapes in laser powder bed fusion for enhanced in situ inspection is proposed to overcome limitations in existing methods concerning photometric variability and domain shift by integrating a Graph Neural Network bottleneck into a U-Net architecture
Motivation and Problem Statement
The technological maturity of in situ inspection and monitoring methods in additive manufacturing is steadily increasing, enabling more efficient and practical qualification procedures In this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been investigated by various authors, leveraging both edge detection and machine learning approaches to identify deviations from nominal geometry Despite these developments, several challenges remain, including the sensitivity of segmentation performance to industrial illumination conditions and layer-to-layer variability in pixel intensity patterns This photometric variability is intrinsic to L-PBF processes and is consistently encountered in real manufacturing scenarios, underscoring the need for robust in situ geometry reconstruction methods
Methodological Approach
The proposed method consists of a neural network architecture that extends the conventional U-Net by integrating a graph-based representation within its bottleneck layer This hybrid architecture, referred to as UNet-GNN, was first introduced by Singh et al. (2025) to improve image segmentation for autonomous driving, where U-Net and its variants struggle to achieve good performance in the presence of highly distorted fisheye images The graph is formulated as G = (V, E), where V denoted nodes and E edges Edges, instead, are created using the k-nearest neighbours (k-NN) formulation, such that each node vx,y connects to the k closest nodes Node features in the t-th layer, namely hx,y t, are updated by aggregating and weighting information from the neighbours of the node in the (t-1)-th layer
Performance Comparison and Robustness
The UNet-GNN accuracy is always higher than the accuracy of competitor methods The gap with respect to both the U-Net and active contour is wider when training and testing conditions are not homogeneous, e.g., training on lattice A and test on lattice B, and viceversa The UNet-GNN is less sensitive to geometry-induced variations from layer to layer as they translate into narrower variations of the segmentation accuracy compared to other two methods Similarly, the UNet-GNN is less sensitive to scan direction-induced variations from layer to layer compared to other two methods In the presence of a gamma perturbation, which affects the global contrast and brightness properties of the image, the performance of both models degrades, but the UNet-GNN remains systematically more accurate than the competitor When Gaussian noise is added to the image, the U-net exhibit an abrupt drop of the segmentation accuracy, while the UNet-GNN yields higher and more stable performance
Computational Efficiency
The active contour methodology exhibits significantly longer inference times, on the order of 16–17 s per image, due to its iterative nature Both U-Net and UNet-GNN models unlock a much faster inference, below 1 s per image Among the two, the UNet-GNN provided the lowest average inference time (about 0.4 s per image) compared to U-Net (about 0. p. 6 1 In situ geometry reconstruction with benchmark methods
In L-PBF, in situ geometry inspection based on post-exposure powder bed images is performed
by acquiring a top-view image of each layer after laser scanning and before powder recoating. At this stage, the solidified regions exhibit distinct optical characteristics compared to the
surrounding loose powder, mainly due to differences in surface morphology and reflectivity.Where the paper says it
Conclusion and Future Work
By integrating a GNN bottleneck within a U-Net architecture, the resulting hybrid architecture enables the explicit modeling of spatial and relational dependencies among image regions, thereby enhancing robustness to the variability inherent to industrial powder bed imaging The experimental results, validated against ground truth data derived from X-ray CT of trabecular lattice structures, demonstrate that UNet-GNN consistently outperforms both active contour methods and standard U-Net architectures across all considered scenarios This robustness represents a key requirement for enabling in situ inspection as a reliable tool to support process qualification and product verification in industrially compliant implementations Future work will focus on exploring the performance of the method in the presence of real defects The integration of graph-based learning within in situ monitoring and in situ inspection frameworks represents a promising direction toward more robust, scalable, and efficient quality assurance practices in AM
Acknowledgements
The authors would like to acknowledge Prof. Bianca Maria Colosimo for coordinating the research activities that contributed to this work The PhD scholarship of Raimondo Stefano is funded by GE Avio S. p. 0 Graph-augmented Segmentation of Complex Shapes in Laser
Powder bed Fusion for Enhanced In Situ Inspection
Stefano Raimondo*
, Matteo Bugatti, Marco Grasso
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa 1, 20156 Milan, Italy
*Corresponding author. raimondo@polimi. This allows enhancing the consistency and accuracy of the geometry reconstruction
in the presence of spatial and layer-wise photometric variability systematically faced in real data. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. This allows enhancing the consistency and accuracy of the geometry reconstruction
in the presence of spatial and layer-wise photometric variability systematically faced in real data. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments.Where the paper says it
Where the paper says it
Where the paper says it
References
Abdelrahman M, Reutzel EW, Nassar AR, Starr TL. (2017). Flaw detection in powder bed fusion using optical imaging. p. 0 The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. ,
2018, Abdelrahman et al. , 2017, Aminzadeh and Kurfess, 2016). The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. Graph-augmented Segmentation of Complex Shapes in Laser
Powder bed Fusion for Enhanced In Situ Inspection
Stefano Raimondo*
, Matteo Bugatti, Marco Grasso
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa 1, 20156 Milan, Italy
*Corresponding author. raimondo@polimi. Graph-augmented Segmentation of Complex Shapes in Laser
Powder bed Fusion for Enhanced In Situ Inspection
Stefano Raimondo*
, Matteo Bugatti, Marco Grasso
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa 1, 20156 Milan, Italy
*Corresponding author. raimondo@polimi. To address this gap, in situ monitoring and inspection approaches leveraging layer-wise data
acquired during fabrication have emerged as enabling technologies and strategic capacities
(Donmez et al. , 2024, ASTM Report, 2023, Fang et al. The alignment between in situ reconstructions and nominal geometries is carried
out during the calibration stage, e. , employing landmark registration methods (more details
can be found in Colosimo et al. Instead, it relies on
supervised training to learn the mapping between input images and segmentation masks,
achieving the desired level of accuracy through data-driven optimization (Ehab et al. , 2024,
Azad et al. Purely local or pixel-driven approaches may
fail to enforce global structural consistency, potentially leading to poor reconstruction
accuracies under challenging imaging conditions. Graph-based segmentation has recently emerged as a promising strategy to address several of
these limitations, with applications largely originating from medical imaging or autonomous
driving domain (Singh et al. A growing body of research has been devoted to this problem (Saini and Shakolas, 2025,
Bugatti et al. 2025, Colosimo et al. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. Keywords: additive manufacturing, image segmentation, graph neural network, in situ inspection, lattice
structure, laser powder bed fusion. Graph-augmented Segmentation of Complex Shapes in Laser
Powder bed Fusion for Enhanced In Situ Inspection
Stefano Raimondo*
, Matteo Bugatti, Marco Grasso
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa 1, 20156 Milan, Italy
*Corresponding author. raimondo@polimi. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In laser powder bed fusion (LPBF), images of the whole powder bed can be captured on a layer-by-layer basis, following
laser exposure, to extract information on the solidification of the geometry at each layer. This
enables the estimation of geometric deviations during the build and allowing potential nonconformities to be detected before part completion, thus moving from post-process, expensive,
and time-consuming inspection to more efficient in situ measurements. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. Through
iterative updates, the contour converges to a configuration that balances geometric regularity
with adherence to background-foreground separation features, resulting in the delineation of
reconstructed boundaries. Over the years, numerous variants have been proposed, evolving from edge-based to regionbased formulations and hybrid models (Chen et al. The performance of the proposed UNet-GNN architecture was benchmarked against two
representative segmentation approaches: active contour methods and conventional U-Net
models. Quantitative validation was conducted using ground-truth data derived from X-ray CT , 2025; Colosimo et al. , 2024, 2022;
Pagani et al. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. Keywords: additive manufacturing, image segmentation, graph neural network, in situ inspection, lattice
structure, laser powder bed fusion. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. Keywords: additive manufacturing, image segmentation, graph neural network, in situ inspection, lattice
structure, laser powder bed fusion. , 2025; Colosimo et al. , 2024, 2022;
Pagani et al. This allows enhancing the consistency and accuracy of the geometry reconstruction
in the presence of spatial and layer-wise photometric variability systematically faced in real data. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. , 2020, Schmitt et al. , 2023,
Dang et al. 1 In situ geometry reconstruction with benchmark methods
In L-PBF, in situ geometry inspection based on post-exposure powder bed images is performed
by acquiring a top-view image of each layer after laser scanning and before powder recoating. At this stage, the solidified regions exhibit distinct optical characteristics compared to the
surrounding loose powder, mainly due to differences in surface morphology and reflectivity. As a result, the gap between technological capability and industrial qualification efficiency
continues to limit the exploitation of metal AM in many production environments. To address this gap, in situ monitoring and inspection approaches leveraging layer-wise data
acquired during fabrication have emerged as enabling technologies and strategic capacities
(Donmez et al. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. To address this gap, in situ monitoring and inspection approaches leveraging layer-wise data
acquired during fabrication have emerged as enabling technologies and strategic capacities
(Donmez et al. , 2024, ASTM Report, 2023, Fang et al. Through
iterative updates, the contour converges to a configuration that balances geometric regularity
with adherence to background-foreground separation features, resulting in the delineation of
reconstructed boundaries. Over the years, numerous variants have been proposed, evolving from edge-based to regionbased formulations and hybrid models (Chen et al. Purely local or pixel-driven approaches may
fail to enforce global structural consistency, potentially leading to poor reconstruction
accuracies under challenging imaging conditions. Graph-based segmentation has recently emerged as a promising strategy to address several of
these limitations, with applications largely originating from medical imaging or autonomous
driving domain (Singh et al. reviewed: active contours (Liu and Peng, 2012; Soomro et al. , 2018) and U-Nets (Ehab et al. segmentation for autonomous driving, where U-Net and its variants struggle to achieve good
performance in the presence of highly distorted fisheye images. In this study, we leverage the
same hybrid architecture for a different scope and in a different domain, showing that
reorganizing intermediate feature representations into a graph structure allows better and more
consistent segmentation results in the presence of natural variations of image properties
commonly faced in L-PBF. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. Graph-augmented Segmentation of Complex Shapes in Laser
Powder bed Fusion for Enhanced In Situ Inspection
Stefano Raimondo*
, Matteo Bugatti, Marco Grasso
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa 1, 20156 Milan, Italy
*Corresponding author. raimondo@polimi. 2025, Colosimo et al. , 2024; Fischer et al. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. Despite these developments, several challenges remain,
including the sensitivity of segmentation performance to industrial illumination conditions and layerto-layer variability in pixel intensity patterns. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. Graph-augmented Segmentation of Complex Shapes in Laser
Powder bed Fusion for Enhanced In Situ Inspection
Stefano Raimondo*
, Matteo Bugatti, Marco Grasso
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa 1, 20156 Milan, Italy
*Corresponding author. raimondo@polimi. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. Graph-augmented Segmentation of Complex Shapes in Laser
Powder bed Fusion for Enhanced In Situ Inspection
Stefano Raimondo*
, Matteo Bugatti, Marco Grasso
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa 1, 20156 Milan, Italy
*Corresponding author. raimondo@polimi. A growing body of research has been devoted to this problem (Saini and Shakolas, 2025,
Bugatti et al. 2025, Colosimo et al. A growing body of research has been devoted to this problem (Saini and Shakolas, 2025,
Bugatti et al. 2025, Colosimo et al. A growing body of research has been devoted to this problem (Saini and Shakolas, 2025,
Bugatti et al. 2025, Colosimo et al. To the best of authors’ knowledge, this is the first study that investigates the potential of hybrid
CNNs and graph neural networks (GNNs) to improve powder bed image segmentation in the
context of in situ monitoring and inspection in L-PBF. The motivating case study regards the in situ inspection of lattice structures, aiming to
anticipate the detection of geometrical deviations and distortions that may have detrimental
effects on the functional performance of the part. To the best of authors’ knowledge, this is the first study that investigates the potential of hybrid
CNNs and graph neural networks (GNNs) to improve powder bed image segmentation in the
context of in situ monitoring and inspection in L-PBF. The motivating case study regards the in situ inspection of lattice structures, aiming to
anticipate the detection of geometrical deviations and distortions that may have detrimental
effects on the functional performance of the part. To the best of authors’ knowledge, this is the first study that investigates the potential of hybrid
CNNs and graph neural networks (GNNs) to improve powder bed image segmentation in the
context of in situ monitoring and inspection in L-PBF. The motivating case study regards the in situ inspection of lattice structures, aiming to
anticipate the detection of geometrical deviations and distortions that may have detrimental
effects on the functional performance of the part. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. Keywords: additive manufacturing, image segmentation, graph neural network, in situ inspection, lattice
structure, laser powder bed fusion. , 2020, Gaikwad et
al. , 2019, He et al. These periodic architectures derive their
functional properties primarily from their geometry rather than from the material itself. As a
result, local geometric inaccuracy, such as strut thinning, missing nodes, or dimensional
distortion, can degrade mechanical, thermal, or functional performance (Helou and Kara, 2018,
Wu et al. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. Keywords: additive manufacturing, image segmentation, graph neural network, in situ inspection, lattice
structure, laser powder bed fusion. , 2024, Liao et al. Originally introduced for biomedical image segmentation,
U-Net has since become a reference architecture in a wide range of computer vision
applications due to its ability to combine precise localization with robust semantic
understanding (Ronneberger et al. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. , 2024, Liao et al. Originally introduced for biomedical image segmentation,
U-Net has since become a reference architecture in a wide range of computer vision
applications due to its ability to combine precise localization with robust semantic
understanding (Ronneberger et al. Moreover, even when properly calibrated, the dependence of active
contour methods on boundary initialization may lead to conservative segmentations biased
toward the nominal geometry, potentially limiting sensitivity to actual deviations. Eventually,
the iterative nature of energy minimization increases computational cost, which restricts realtime applicability, particularly when high-resolution powder bed images are processed at
industrial scale. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. Keywords: additive manufacturing, image segmentation, graph neural network, in situ inspection, lattice
structure, laser powder bed fusion. , 2024,
Azad et al. , 2024, Liao et al. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. reviewed: active contours (Liu and Peng, 2012; Soomro et al. , 2018) and U-Nets (Ehab et al. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. reviewed: active contours (Liu and Peng, 2012; Soomro et al. , 2018) and U-Nets (Ehab et al. This graph-based representation allows the model to “capture” the underlying geometry
encoded in the image even when photometric properties deviate from the ones observed during
the training and vary across the build area or from layer to layer. By decoupling structural
consistency from local intensity fluctuations, the approach enhances robustness to spatial and
layer-wise photometric variability while enhancing segmentation accuracy and topological
consistency. Through
iterative updates, the contour converges to a configuration that balances geometric regularity
with adherence to background-foreground separation features, resulting in the delineation of
reconstructed boundaries. Over the years, numerous variants have been proposed, evolving from edge-based to regionbased formulations and hybrid models (Chen et al. Eventually,
the iterative nature of energy minimization increases computational cost, which restricts realtime applicability, particularly when high-resolution powder bed images are processed at
industrial scale. More recently, convolutional neural network (CNN) models, and in particular U-Net
architectures, have demonstrated notable performance as well as enhanced computational
efficiency for in situ geometry reconstruction in L-PBF (Scime et al. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. , 2017; Li et al. These
approaches iteratively refine an initial boundary through shrink/expand operations driven by
the minimization of an energy functional, typically implemented within a level set framework
(Osher et al. (2022b) and Colosimo et al. However, it is worth
noticing the proposed approach is not constrained to one specific geometry, and it can be
generally applied to every kind of manufactured shape. To the best of authors’ knowledge, this is the first study that investigates the potential of hybrid
CNNs and graph neural networks (GNNs) to improve powder bed image segmentation in the
context of in situ monitoring and inspection in L-PBF. The motivating case study regards the in situ inspection of lattice structures, aiming to
anticipate the detection of geometrical deviations and distortions that may have detrimental
effects on the functional performance of the part. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. , 2024, Liao et al. Originally introduced for biomedical image segmentation,
U-Net has since become a reference architecture in a wide range of computer vision
applications due to its ability to combine precise localization with robust semantic
understanding (Ronneberger et al. , 2024, Liao et al. Originally introduced for biomedical image segmentation,
U-Net has since become a reference architecture in a wide range of computer vision
applications due to its ability to combine precise localization with robust semantic
understanding (Ronneberger et al. This
enables the estimation of geometric deviations during the build and allowing potential nonconformities to be detected before part completion, thus moving from post-process, expensive,
and time-consuming inspection to more efficient in situ measurements. A growing body of research has been devoted to this problem (Saini and Shakolas, 2025,
Bugatti et al. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. , 2025, Lu et al. 2020, Scarselli et al. The study addresses these limitations by proposing a
graph-augmented segmentation approach. The underlying principle consists of preserving the
geometrical information at a global level rather than at pixel-wise level, modeling dependencies and
relational information among spatial regions with a Graph Neural Network bottleneck embedded into a
U-Net architecture. As a further effect, it maintains or even improves the computational efficiency,
thereby enabling real-time implementation under stringent latency constraints. To the best of authors’ knowledge, this is the first study that investigates the potential of hybrid
CNNs and graph neural networks (GNNs) to improve powder bed image segmentation in the
context of in situ monitoring and inspection in L-PBF. , 2017, Aminzadeh and Kurfess, 2016). The underlying idea consists
of reconstructing the contours of the solidified layer via powder bed image segmentation and
comparing it against the nominal shape to detect anomalous deviations. Eventually,
the iterative nature of energy minimization increases computational cost, which restricts realtime applicability, particularly when high-resolution powder bed images are processed at
industrial scale. More recently, convolutional neural network (CNN) models, and in particular U-Net
architectures, have demonstrated notable performance as well as enhanced computational
efficiency for in situ geometry reconstruction in L-PBF (Scime et al. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. segmentation for autonomous driving, where U-Net and its variants struggle to achieve good
performance in the presence of highly distorted fisheye images. In this study, we leverage the
same hybrid architecture for a different scope and in a different domain, showing that
reorganizing intermediate feature representations into a graph structure allows better and more
consistent segmentation results in the presence of natural variations of image properties
commonly faced in L-PBF. To overcome these limitations, we propose moving from traditional convolutional backbones
(e. , U-Nets) to a “graph-augmented segmentation” approach designed to preserve the
geometrical information at a global level rather than at a pixel-level, by modeling dependencies
and relational information among spatial regions (Singh et al. In this study, we leverage the
same hybrid architecture for a different scope and in a different domain, showing that
reorganizing intermediate feature representations into a graph structure allows better and more
consistent segmentation results in the presence of natural variations of image properties
commonly faced in L-PBF. This graph-based representation allows the model to “capture” the underlying geometry
encoded in the image even when photometric properties deviate from the ones observed during
the training and vary across the build area or from layer to layer. In situ deviation detection is then
achieved by comparing the reconstructed geometry against the nominal slice derived from the
CAD model. The alignment between in situ reconstructions and nominal geometries is carried
out during the calibration stage, e. , 2024, Liao et al. Originally introduced for biomedical image segmentation,
U-Net has since become a reference architecture in a wide range of computer vision
applications due to its ability to combine precise localization with robust semantic
understanding (Ronneberger et al. , 2024, Liao et al. Originally introduced for biomedical image segmentation,
U-Net has since become a reference architecture in a wide range of computer vision
applications due to its ability to combine precise localization with robust semantic
understanding (Ronneberger et al. Moreover, even when properly calibrated, the dependence of active
contour methods on boundary initialization may lead to conservative segmentations biased
toward the nominal geometry, potentially limiting sensitivity to actual deviations. Eventually,
the iterative nature of energy minimization increases computational cost, which restricts realtime applicability, particularly when high-resolution powder bed images are processed at
industrial scale. Moreover, even when properly calibrated, the dependence of active
contour methods on boundary initialization may lead to conservative segmentations biased
toward the nominal geometry, potentially limiting sensitivity to actual deviations. Eventually,
the iterative nature of energy minimization increases computational cost, which restricts realtime applicability, particularly when high-resolution powder bed images are processed at
industrial scale. As a
result, local geometric inaccuracy, such as strut thinning, missing nodes, or dimensional
distortion, can degrade mechanical, thermal, or functional performance (Helou and Kara, 2018,
Wu et al. , 2019, Tang et al. The
method is evaluated against benchmark techniques for the in situ reconstruction of lattice structures
produced by L-PBF, demonstrating its potential as a scalable solution for robust in situ inspection and
geometric verification in industrial environments. Keywords: additive manufacturing, image segmentation, graph neural network, in situ inspection, lattice
structure, laser powder bed fusion. Graph-augmented Segmentation of Complex Shapes in Laser
Powder bed Fusion for Enhanced In Situ Inspection
Stefano Raimondo*
, Matteo Bugatti, Marco Grasso
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa 1, 20156 Milan, Italy
*Corresponding author. raimondo@polimi. Introduction
Additive Manufacturing (AM) has reached technological maturity across different industrial
sectors, including aerospace, energy, biomedical, and creative industries. This maturity has
been achieved through continuous improvements in process quality and system reliability, the
integration of AM technologies into advanced production lines, increasing levels of
automation, and continuous improvement in material development. , U-Nets) to a “graph-augmented segmentation” approach designed to preserve the
geometrical information at a global level rather than at a pixel-level, by modeling dependencies
and relational information among spatial regions (Singh et al. , 2025, Xiao et al. As a further effect, it maintains or even improves the computational efficiency,
thereby enabling real-time implementation under stringent latency constraints. To the best of authors’ knowledge, this is the first study that investigates the potential of hybrid
CNNs and graph neural networks (GNNs) to improve powder bed image segmentation in the
context of in situ monitoring and inspection in L-PBF. (2021), McCann et al. (2021), Zhang and Yan (2023), Fang et al. In
this context, image segmentation of powder bed images in Laser Powder Bed Fusion (L-PBF) has been
investigated by various authors, leveraging both edge detection and machine learning approaches to
identify deviations from nominal geometry. Despite these developments, several challenges remain,
including the sensitivity of segmentation performance to industrial illumination conditions and layerto-layer variability in pixel intensity patterns. , 2019, He et al. , 2019, Zur Jacobsmühlen et al. In laser powder bed fusion (LPBF), images of the whole powder bed can be captured on a layer-by-layer basis, following
laser exposure, to extract information on the solidification of the geometry at each layer. This
enables the estimation of geometric deviations during the build and allowing potential nonconformities to be detected before part completion, thus moving from post-process, expensive,
and time-consuming inspection to more efficient in situ measurements. it
Abstract. The technological maturity of in situ inspection and monitoring methods in additive
manufacturing is steadily increasing, enabling more efficient and practical qualification procedures. By repeating the comparison between the reconstructed
geometry and the nominal one layer by layer, it becomes possible to detect dimensional
inaccuracies, incomplete fusion, local swelling, recoater-induced defects affecting the part
consolidation, or topological discontinuities. For a review of these methods, the reader is
referred to Grasso et al. By repeating the comparison between the reconstructed
geometry and the nominal one layer by layer, it becomes possible to detect dimensional
inaccuracies, incomplete fusion, local swelling, recoater-induced defects affecting the part
consolidation, or topological discontinuities. For a review of these methods, the reader is
referred to Grasso et al.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Table Summary
Table 1 summarizes the main differences between the standard U-Net architecture and the proposed UNet-GNN Table 4 Part I – Mean accuracy and corresponding 95% confidence intervals for the UNet-GNN and the U-Net evaluated under different image data perturbation with different severity levels. Results are reported separately for nodes (a) and struts (b) Table 4 Part II – Mean accuracy and corresponding 95% confidence intervals for the UNet-GNN and the U-Net evaluated under different image data perturbation with different severity levels. p. 6 To further evaluate robustness under realistic sources of
image variability, all methods were tested under controlled perturbation scenarios, including
additive Gaussian noise, gamma-based intensity transformations, and progressive pixelation at
increasing severity levels. The results show that UNet-GNN consistently outperforms the
benchmark approaches, making segmentation accuracy less sensitive to different sources of
undesired variability. To further evaluate robustness under realistic sources of
image variability, all methods were tested under controlled perturbation scenarios, including
additive Gaussian noise, gamma-based intensity transformations, and progressive pixelation at
increasing severity levels. The results show that UNet-GNN consistently outperforms the
benchmark approaches, making segmentation accuracy less sensitive to different sources of
undesired variability. To further evaluate robustness under realistic sources of
image variability, all methods were tested under controlled perturbation scenarios, including
additive Gaussian noise, gamma-based intensity transformations, and progressive pixelation at
increasing severity levels. The results show that UNet-GNN consistently outperforms the
benchmark approaches, making segmentation accuracy less sensitive to different sources of
undesired variability. To further evaluate robustness under realistic sources of
image variability, all methods were tested under controlled perturbation scenarios, including
additive Gaussian noise, gamma-based intensity transformations, and progressive pixelation at
increasing severity levels. The results show that UNet-GNN consistently outperforms the
benchmark approaches, making segmentation accuracy less sensitive to different sources of
undesired variability.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Graph-augmented segmentation improves robustness to photometric variability by
decoupling structural consistency from local intensity fluctuations,
which allows it toenhance robustness to spatial and layer-wise photometric variability while enhancing segmentation accuracy.
-
The UNet-GNN architecture can capture global geometric context and relational dependencies, enabling the model to learn not only
local appearance cues but also higher-order connectivity patterns and structural constraints.
-
The improved system can perform in situ geometry reconstruction with enhanced reliability across different build locations and layers, as demonstrated by the finding that
the UNet-GNN consistently outperforms both active contour methods and standard U-Net architectures across all considered scenarios.
-
The model exhibits reduced sensitivity to geometry-induced variations from layer to layer and scan direction variations, which are
intrinsic to L-PBF processes,
leading to more consistent performance across varying acquisition conditions. -
The system maintains fast inference times comparable to or better than standard U-Net models (about 0.4 s per image), making it suitable for real-time implementation in AM where
computational responsiveness is as critical as segmentation accuracy.
Abstract
In situ geometric inspection in Laser Powder Bed Fusion of Metals (PBF-LB/M) opens new possibilities for anticipated defect detection and more efficient qualification. It relies on image segmentation to reconstruct layer-wise geometries and identify manufacturing deviations as the part is being built. However, existing approaches face a critical yet largely overlooked challenge: photometric out-of-distribution (OOD) generalization, namely the ability to maintain reconstruction accuracy when image appearance differs from the conditions represented during model training. This study represents the first exploration of a graph-augmented U-Net architecture for in situ geometric inspection in PBF-LB/M. The hybrid network integrates a Graph Neural Network (GNN) within the bottleneck to enhance reconstruction robustness under photometric distribution shifts. It is evaluated against conventional U-Net and active contour segmentation methods using two experimental case studies involving different lattice geometries, materials, and manufacturing setups, with X-ray computed tomography providing reference geometries. Generalization is assessed across different training and inference conditions and under controlled image perturbations. Results show that the UNet-GNN achieves higher reconstruction accuracy and consistency than the benchmark methods, with more pronounced improvements when training and inference conditions differ. Its advantages are confirmed across both case studies and under controlled photometric perturbations, while preliminary anomaly detection experiments indicate that improved robustness does not necessarily compromise sensitivity to geometric deviations. These findings highlight graph augmentation as a promising strategy for more reliable in situ geometric inspection under heterogeneous manufacturing conditions.
Sources
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs