mlsanityData to 7 Oct 2026

SYNOSIS: Image synthesis pipeline for machine vision in metal surface inspection

arXiv 2410.14844 ↗First posted 18 Oct 2024cs.CV cs.CE cs.GRcited by 2

Not on the map. Why?

Abstract (arXiv)

The use of machine learning (ML) methods for development of robust and flexible visual inspection system has shown promising. However their performance is highly dependent on the amount and diversity of training data. This is often restricted not only due to costs but also due to a wide variety of defects and product surfaces which occur with varying frequency. As such, one can not guarantee that the acquired dataset contains enough defect and product surface occurrences which are needed to develop a robust model. Using parametric synthetic dataset generation, it is possible to avoid these issues. In this work, we introduce a complete pipeline which describes in detail how to approach image synthesis for surface inspection - from first acquisition, to texture and defect modeling, data generation, comparison to real data and finally use of the synthetic data to train a defect segmentation model. The pipeline is in detail evaluated for milled and sandblasted aluminum surfaces. In addition to providing an in-depth view into each step, discussion of chosen methods, and presentation of ML results, we provide a comprehensive dual dataset containing both real and synthetic images.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 34 of its 127 references that are in the map. First posted on arXiv 18 Oct 2024.

Builds on

34 of 127 references.

References · 34

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  3. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,029
  4. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,857
  5. 2017CARLA: An Open Urban Driving Simulatorcited by 2,431
  6. 2022Kubric: A scalable dataset generatorcited by 373
  7. 2017Learning from Synthetic Humanscited by 358
  8. 2016Virtual Worlds as Proxy for Multi-Object Tracking Analysiscited by 340
  9. 2018Training Deep Networks with Synthetic Data: Bridging the Reality Gap by Domain Randomizationcited by 292
  10. 2020A Brief Review of Domain Adaptationcited by 287
  11. 2018Deep Object Pose Estimation for Semantic Robotic Grasping of Household Objectscited by 237
  12. 2021VT-ADL: A Vision Transformer Network for Image Anomaly Detection and Localizationcited by 222
22 more
  1. 2021Fake It Till You Make It: Face analysis in the wild using synthetic data alonecited by 200
  2. 2019Segmentation-Based Deep-Learning Approach for Surface-Defect Detectioncited by 145
  3. 2021Mixed supervision for surface-defect detection: from weakly to fully supervised learningcited by 101
  4. 2018Learning to Detect and Track Visible and Occluded Body Joints in a Virtual Worldcited by 81
  5. 2021Defect-GAN: High-Fidelity Defect Synthesis for Automated Defect Inspectioncited by 77
  6. 2017Learning Texture Manifolds with the Periodic Spatial GANcited by 53
  7. 2016Texture Synthesis with Spatial Generative Adversarial Networkscited by 45
  8. 2019Precise Synthetic Image and LiDAR (PreSIL) Dataset for Autonomous Vehicle Perceptioncited by 29
  9. 2021An Inverse Procedural Modeling Pipeline for SVBRDF Mapscited by 24
  10. 2018Training Deep Face Recognition Systems with Synthetic Datacited by 22
  11. 2023Real-Time Neural Appearance Modelscited by 21
  12. 2022MIAD: A Maintenance Inspection Dataset for Unsupervised Anomaly Detectioncited by 19
  13. 2016Play and Learn: Using Video Games to Train Computer Vision Modelscited by 16
  14. 2017A survey of exemplar-based texture synthesiscited by 13
  15. 2022CAD2Render: A Modular Toolkit for GPU-accelerated Photorealistic Synthetic Data Generation for the Manufacturing Industrycited by 10
  16. 2023Defect Transfer GAN: Diverse Defect Synthesis for Data Augmentationcited by 9
  17. 2022Dataset of Industrial Metal Objectscited by 8
  18. 2024Towards Sim-to-Real Industrial Parts Classification with Synthetic Datasetcited by 7
  19. 2024The Effects of Hallucinations in Synthetic Training Data for Relation Extractioncited by 7
  20. 2021Industrial Machine Tool Component Surface Defect Datasetcited by 5
  21. 2021Rail-5k: a Real-World Dataset for Rail Surface Defects Detectioncited by 4
  22. 2024Stochastic Geometry Models for Texture Synthesis of Machined Metallic Surfaces: Sandblasting and Millingcited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026SynthRender and I-AsSET: Open-Source Framework and Dataset for Bidirectional Sim-Real Transfer in Industrial Object Perceptioncited by 1
  2. 2026Synthetic Defect Geometries of Cast Metal Objects Modeled via 2d Voronoi Tessellationscited by 0