PerceptionANS

Trench_stones

Instance Segmentation

Trench_stones Image Dataset

v1

2024-02-27 5:07pm

Generated on Feb 27, 2024

Dataset Split

Train Set 88%
1239Images
Valid Set 8%
118Images
Test Set 4%
59Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 3
Flip: Horizontal
Crop: 0% Minimum Zoom, 20% Maximum Zoom
Shear: ±10° Horizontal, ±10° Vertical
Hue: Between -15° and +15°
Saturation: Between -58% and +58%
Brightness: Between -49% and +49%
Exposure: Between -11% and +11%
Blur: Up to 2.5px
Noise: Up to 0.1% of pixels
Cutout: 15 boxes with 1% size each