Spectral Labs Marine Dataset - Segmented Dataset
Versions
2023-07-17 9:42pm
v11
· a year ago
2023-07-16 8:16pm
v10
· a year ago
2023-07-16 8:04pm
v9
· a year ago
2023-07-16 3:32pm
v8
· a year ago
2023-07-16 2:55pm
v7
· a year ago
2023-07-16 9:18am
v6
· a year ago
2023-07-15 3:30pm
v5
· a year ago
2023-07-15 11:16am
v4
· a year ago
2023-07-15 11:03am
v3
· a year ago
2023-07-14 6:57pm
v2
· a year ago
2023-07-14 6:52pm
v1
· a year ago
v6
2023-07-16 9:18am
Generated on Jul 16, 2023
3553 Total Images
View All ImagesDataset Split
Train Set 86%
3058Images
Valid Set 13%
448Images
Test Set 1%
47Images
Preprocessing
Resize: Fit (black edges) in 640x640
Modify Classes: 36 remapped, 1 dropped
Filter Null: Require all images to contain annotations.
Augmentations
Outputs per training example: 2
Flip: Horizontal, Vertical
Crop: 0% Minimum Zoom, 20% Maximum Zoom
Shear: ±15° Horizontal, ±15° Vertical
Hue: Between -30° and +30°
Saturation: Between -40% and +40%
Brightness: Between -30% and +30%
Blur: Up to 2.5px
Noise: Up to 10% of pixels
Cutout: 25 boxes with 10% size each