Universidad Autonoma de Nuevo Leon

perka

Object Detection

perka Image Dataset

v5

2023-05-15 1:54am

Generated on May 15, 2023

Popular Download Formats

Dataset Split

Train Set 67%
6518Images
Valid Set 10%
1008Images
Test Set 23%
2257Images

Preprocessing

Auto-Orient: Applied
Isolate Objects: Applied
Static Crop: 25-75% Horizontal Region, 25-75% Vertical Region
Resize: Stretch to 640x640
Auto-Adjust Contrast: Using Adaptive Equalization
Grayscale: Applied
Modify Classes: 7 remapped, 1 dropped

Augmentations

Outputs per training example: 2
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise
Rotation: Between -15° and +15°
Blur: Up to 4.75px
Bounding Box: Flip: Horizontal
Bounding Box: 90° Rotate: Clockwise, Counter-Clockwise
Bounding Box: Rotation: Between -15° and +15°
Bounding Box: Blur: Up to 2.5px