beric7 / structural_inspection_main

Collection of works in my PhD at Virginia Tech, mostly focused on computer vision and machine learning applications in structural inspection and structural health monitoring.

Structural Inspection Main

Updates!

• 1/22/2024 - Code for Registration using Superglue and RANSACFlow
• 8/7/2023 - Updated links and changed a lot of red-circles to green-circles (in other words, the links to papers, datasets, etc. are updated).
• 5/17/2023 - Its been a while... but I am planning to make big updates and refactor structural inspection main (stay tuned!)
• 12/14/2021 - Annotation study repository has been published.
• 12/31/2021 - StyleGAN2 repository has been published.

Associated Papers and Repositories

"Visual Structural Inspection Datasets" - 2022 (E. Bianchi, M. Hebdon) 🟢 [Paper]

Link to access the table, contribute dataset links, submit edit request, or become a data steward:

• 🟢 Cataloged Dataset

---

"Development of Open-source Collaborative Structural Inspection Datasets" - 2022 (E. Bianchi, M. Hebdon) 🟢 [Paper]

Access all six datasets, the code, and the trained models here:

• 🟢 Development of Open-source Collaborative Structural Inspection Datasets

---

"Forecasting Infrastructure Deterioration with GAN-Inversion" - 2021 (E. Bianchi, M. Hebdon) 🟢 [Paper] 🟢[Open Access Paper]

• 🟢 GAN-Inversion codebase: StyleGAN2 and InterfaceGAN implementation

---

"Bridge Inspection Component Registration for Damage Evolution" - 2022 (E. Bianchi, S. Nazmus, C. Woolsey, M. Hebdon) 🟢 [Paper]

• 🟢 Component Registration codebase and results

Note the Referenced Papers

SuperGlue paper, we based our code on the official SuperGlue repository
RANSAC-FLOW paper, we based our code on the official RANSAC-FLOW repository

---

"Extending Datasets with Editable First-pass Annotations" - 2021 (E. Bianchi) 🟢 [Paper]

• 🟢 Editable First-pass Annotations codebase and results

---

structural_inspection_main's Issues

Dataset catalog table link not working

Hello, thanks for the work gathering all these datasets!
The link to the google sheet table is no longer working, could you please check it? Thanks!

Can't find source code for DeeplabV3+ training

Where can I find source code for training DeeplabV3+ on corrosion segmentation task? When I am trying to load provided weights:

state_dict = torch.load('l2_loss/weights_35.pt')

I got the following error:

ModuleNotFoundError: No module named 'network'