MS COCO Dataset

MS COCO Object Detector

Object Detection

MS COCO Object Detector Image Dataset

v1

2023-05-09 11:05pm

Generated on May 10, 2023

Popular Download Formats

Dataset Split

Train Set 97%
3141Images
Valid Set 2%
55Images
Test Set 2%
54Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 1200x1200

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Crop: 0% Minimum Zoom, 55% Maximum Zoom
Rotation: Between -21° and +21°
Shear: ±10° Horizontal, ±29° Vertical
Grayscale: Apply to 23% of images
Hue: Between -17° and +17°
Saturation: Between -32% and +32%
Brightness: Between -42% and +42%
Exposure: Between -43% and +43%
Blur: Up to 3px
Noise: Up to 5% of pixels
Cutout: 3 boxes with 16% size each
Mosaic: Applied
Bounding Box: Flip: Horizontal, Vertical
Bounding Box: 90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Bounding Box: Crop: 0% Minimum Zoom, 20% Maximum Zoom
Bounding Box: Rotation: Between -19° and +19°
Bounding Box: Shear: ±15° Horizontal, ±15° Vertical
Bounding Box: Brightness: Between -25% and +25%
Bounding Box: Exposure: Between -24% and +24%
Bounding Box: Blur: Up to 4.5px
Bounding Box: Noise: Up to 2% of pixels