The SKU110k dataset includes 11,762 images taken in dense scenes, with more than 1.7 million bounding box annotations and approximately 1,733,678 instances. The dataset includes:

• Training set: 8,233 images
• Validation set: 588 images
• Test set: 2,941 images These images are collected from thousands of supermarket stores with different scales, viewpoints, lighting conditions, and noise levels. All images are resized to a resolution of one million pixels. Most instances in the dataset are closely packed and are usually tilted in the range of [-15∘, 15∘].

Citation

[1] Eran Goldman*, Roei Herzig*, Aviv Eisenschtat*, Jacob Goldberger, Tal Hassner, Precise Detection in Densely Packed Scenes, 2019. [2] Tsung-Yi Lin, Priyal Goyal, Ross Girshick, Kaiming He, Piotr Dollar, Focal loss for dense object detection, 2018.