-
Notifications
You must be signed in to change notification settings - Fork 11
Datasets
Gustavo Rosa edited this page Mar 15, 2017
·
2 revisions
In directory LibOPF/data, the user can find several artificial and synthetic datasets, which are divided into:
Synthetic:
- Datasets obtained from [Kuncheva]:
- cone-torus.dat: dataset containg 400 samples distributed into 3 classes. Each sample is represented by 2 features.
- petals.dat: dataset containg 100 samples distributed into 4 classes. Each sample is represented by 2 features.
- boat.dat: dataset containing 100 samples distributed into 3 classes. Each sample is represented by 2 features.
- saturn.dat: dataset containing 200 samples distributed into 2 classes. Each sample is represented by 2 features.
Artificial:
-
MPEG-7 shape dataset [MPEG-7]. The MPEG-7 shape dataset contains 1400 samples distributed into 70 classes:
- mpeg7_FOURIER.dat: Fourier shape descritor extracted from the whole dataset (126 features).
- mpeg7_BAS.dat: BAS (Beam Angle Statistics) shape descritor extracted from the whole dataset (180 features).
-
Details about the features extraction procedure can be found in [PapaIJIST09].