Skip to content

Latest commit

 

History

History
29 lines (23 loc) · 1.15 KB

File metadata and controls

29 lines (23 loc) · 1.15 KB

1. Create base images (Google Earth)

2. Create base models (Google Earth)

3. Prepare WikiScenes exterior images

  1. Create an empty directory for collected images.
  2. Run py collect_exterior_images.py --input_dir <path to wikiscenes 'cathedrals' dir> --output_dir <path to your collected photos dir>
  3. Create an empty directory for renamed images.
  4. Run rename_images_ascii.py --input_dir <path to your collected photos dir> --output_dir <path to your renamed photos dir>

4. Create extended models

## 5. Search for number of inliers:
1. Create base-ext databases.
2. Analyze cross-dataset matches and get the inlier search bounds.
3. Create cloud instances. (?)
4. Iterate inliers.
	- Challenge: unintentionally creating multiple base models.
5. Plot graphs of # of images vs. # of inliers to find
6. iterate over candidates and create full reconstructions.
7. Analyze candidate reconstructions qualitatively & quantitatively.
8. Upload chosen models. (?)

## 6. Optimal - Search across or refine other parameters if needed. 

5. Post-process/Analysis:

  • Normalize models.
  • Create scoresheets.
  • Save ext-only reconstruction (optional)