Purpose: simplify sorting of big photo catalog
Stages:
- Detect similar photos
- Select best photo among similar photos (according to different score and weights per computer)
- compute sharpest quality
- compute largest scene
- any other computer weights
- Select the best photo (if no ambiguity) or a subset of best photos
- Outputs will be a text file with columns:
photo-path
group-id (identify the similar photos)
score-total (taking in account weight parameters)
score-<computer-name> raw result of each computer (allow to avoid computing again if value exists)