https://github.com/comnik/missing-labels
Final project for the LIST course at ETH.
https://github.com/comnik/missing-labels
Last synced: 10 months ago
JSON representation
Final project for the LIST course at ETH.
- Host: GitHub
- URL: https://github.com/comnik/missing-labels
- Owner: comnik
- Created: 2015-05-30T14:53:26.000Z (about 11 years ago)
- Default Branch: master
- Last Pushed: 2015-05-31T13:43:05.000Z (about 11 years ago)
- Last Synced: 2025-08-23T09:48:11.279Z (10 months ago)
- Language: Python
- Size: 789 KB
- Stars: 0
- Watchers: 4
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Semi-supervised Image Classification
## Usage
General:
python classify.py [caching options] [output mode]
By default, everything is loaded from snapshots in the snapshot/ subdirectory.
Caching options are **--reload** and **--reclassify**. Those trigger recomputing
input features and labels (and snapshotting them) or re-training the classifier respectiveley.
Output modes are **--cv** for cross-validation, **--validate** for generating validation-set predictions
and finally **--test** for generating test-set predictions. Predictions are stored in the out/ subdirectory.
So when executing classify.py for the first time:
python classify.py --reload --reclassify --cv
Afterwards you can usually stop reloading:
python classify.py --reclassify [output mode]
After finding the best classifier, you can quickly generate different outputs:
python classify.py --cv
python classify.py --validate
python classify.py --test