Update docs: target-size 512, dataset generated

- Drone images now 512x512 (not 256x256)
- Satellite crops saved at native 512x512 (no downscale)
- Dataset generated: 25 GB on disk
- Added known issue: 6 drones in route 06 outside satellite coverage

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
pikaliov
2026-04-18 02:41:29 +03:00
parent fd014a3155
commit cf455cd0f3
3 changed files with 23 additions and 6 deletions

View File

@@ -9,8 +9,8 @@ compatible with UAV-GeoLoc format.
```
UAV_VisLoc_dataset/ UAV_VisLoc_processed/
├── 01/ ├── 01/
│ ├── drone/*.JPG (3976x2652) ---> │ ├── drone/*.JPG (256x256)
│ ├── satellite01.tif ---> │ ├── DB/img/crop_X_Y.png (256x256)
│ ├── drone/*.JPG (3976x2652) ---> │ ├── drone/*.JPG (512x512)
│ ├── satellite01.tif ---> │ ├── DB/img/crop_X_Y.png (512x512)
│ └── 01.csv │ ├── DB/db_postion.txt
│ │ ├── positive.json
├── ... │ └── semi_positive.json
@@ -43,9 +43,9 @@ python scripts/prepare_dataset.py --src ... --dst ... --routes 01 02 03
## Steps
1. **Resize drone images** -> 256x256 JPEG (quality=95)
1. **Resize drone images** -> 512x512 JPEG (quality=95)
2. **Stitch satellite tiles** for route 09 (4 tiles -> 44800x33280)
3. **Crop satellite maps** -> 512x512 patches, stride 256 (50% overlap), resize -> 256x256 PNG
3. **Crop satellite maps** -> 512x512 patches, stride 256 (50% overlap), saved as 512x512 PNG
4. **Compute GPS** for each crop center from satellite bbox + grid position
5. **Match drone -> crops** via vectorized haversine (positive = closest, semi-positive = +-1 in grid)
6. **Write metadata**: positive.json, semi_positive.json, db_postion.txt (per route)
@@ -107,6 +107,16 @@ Columns: name, longitude, latitude, scale_lon (deg/px), scale_lat (deg/px).
| **Total** | **6,744** | **74,807** | | |
Split: 5,060 train / 1,684 test queries. Gallery: 74,807 crops (shared).
Disk size: **25 GB**.
## Image Resolution
All images stored at **512x512** on disk:
- Drone: resized from 3976x2652 / 3000x2000 -> 512x512
- Satellite crops: cut at 512x512 from satellite map, no downscale
Resolution 512 chosen to support downstream tasks (segmentation, depth, normals, canopy height).
Resize to 224/256 for model input should be done in the dataloader, not on disk.
## GPS Matching Quality