Summary of Integrating Artificial Intelligence Models and Synthetic Image Data For Enhanced Asset Inspection and Defect Identification, by Reddy Mandati et al.
Integrating Artificial Intelligence Models and Synthetic Image Data for Enhanced Asset Inspection and Defect Identification
by Reddy Mandati, Vladyslav Anderson, Po-chen Chen, Ankush Agarwal, Tatjana Dokic, David Barnard, Michael Finn, Jesse Cromer, Andrew Mccauley, Clay Tutaj, Neha Dave, Bobby Besharati, Jamie Barnett, Timothy Krall
First submitted to arxiv on: 15 Oct 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
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 The proposed solution combines synthetic asset defect images with manually labeled drone images to improve the performance of automated defect detection, reduce manual labeling hours, and enable realistic image generation for rare defects. A workflow combining 3D modeling tools like Maya and Unreal Engine is used to create photorealistic 2D renderings and 3D models of defective assets. These synthetic images are integrated into a training pipeline augmenting real data. The study implements an end-to-end AI solution for detecting assets and asset defects from combined imagery. The unique contribution lies in applying advanced computer vision models and generating photorealistic 3D renderings. The asset detection model achieved an accuracy of 92%, with a performance lift of 67% when introducing approximately 2,000 synthetic images of 2k resolution. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Drone-based inspections are helping utilities identify asset defects more efficiently. However, using this data for automated defect detection requires a lot of manual labeling work. Researchers have come up with a new way to make this process better by combining real and fake images. They used special computer tools like Maya and Unreal Engine to create realistic 2D and 3D images of defective assets. These synthetic images were then mixed with the real data to train an AI model that can detect assets and defects. The results showed that the AI model was very accurate, achieving a score of 92%. This new approach could make it easier for utilities to maintain their infrastructure. |
Keywords
» Artificial intelligence » Image generation
