Multi-illumination is a highly sought-after topic for training machine learning models. That said, it’s not always easy to find multi-illumination datasets to train your models.
That’s why we’ve done the tricky bit for you. We’ve searched high and low here at Twine to find the best multi-illumination datasets.
Are you ready?
Let’s dive in.
Here are our top picks for Multi-Illumination Datasets:
Multi-Illumination Images in the Wild Dataset
This dataset contains more than 1000 real scenes, each captured under 25 lighting conditions. These models are trained for three challenging applications: single-image illumination estimation, image relighting, and mixed-illuminant white balance.
Large Scale Multi-Illuminant (LSMI) Dataset
This dataset contains 7,486 images, captured with three different cameras on more than 2,700 scenes with two or three illuminants. For each image in the dataset, the pixel-wise ground truth illumination is provided, as well as the chromaticity of each illuminant in the scene and the mixture ratio of illuminants per pixel. Images in this dataset are mostly captured with illuminants existing in the scene, and the ground truth illumination is computed by taking the difference between the images with different illumination combinations.
CSEM’s Multi-Illumination Surface Defect Detection Dataset
The dataset consists of three different types of metallic parts — washers, screws, and gears (temporarily only the first one is available). Parts were captured in a half-spherical light-dome system that filtered out all the ambient light and successively illuminated it from 108 distinct illumination angles. Each 12 illumination angles share the same elevation level and the relative azimuthal difference between the adjacent illumination angles on the same level is 30 degrees. This dataset contains 32 samples, and the test set 38 samples. Each sample comprises 108 images (each captured under a different illumination angle), an automatically extracted foreground segmentation mask, and a hand-labeled defect segmentation mask.
Varying Illumination Dataset
This dataset was collected to evaluate multiple view change detection algorithms under greatly varying illumination. Thousands of pictures were taken at a Brown University quadrangle, the Lincoln Field, three times a day and three times a week for 6 months, between April and November 2006. The scene was observed from roughly 6 viewpoints and its geometry is composed of a low curvature ground (grass and concrete), buildings, trees, bushes, and lampposts. Dynamic foreground objects include vehicles, trash cans, people, objects, etc.
Many illumination changes are observed due to Sun’s position, clouds, overcast weather, and rain. This phenomenon affects the shadows’ location, relative intensity, and also color.
Flash and Ambient Illumination Pairs Dataset
This dataset covers a wide variety of scenes captured by many casual photographers. It explores the computational effort needed to generate completely separate flash illuminations from the ambient light in an uncontrolled setup. The photographs were captured in 12-bit raw and jpeg formats using iPhone 6s and iPhone 7 devices. The flash and no-flash photograph pairs were captured in bursts, typically 0.5-1.0 seconds apart. The two photographs were then aligned via homography estimations, cropped, and resized to 1080×1440 resolution.
To conclude, here are the top picks for the best multi-illumination datasets for your projects:
- Multi-Illumination Images in the Wild Dataset
- Large Scale Multi-Illuminant (LSMI) Dataset
- CSEM’s Multi-Illumination Surface Defect Detection Dataset
- Varying Illumination Dataset
- Flash and Ambient Illumination Pairs Dataset
We hope that this list has helped you find a dataset for your project or, realize the myriad options available.
Please let us know if there are any datasets you would like us to add to the list.
If you want to learn more about how we could help build a custom dataset for your project, don’t hesitate to contact us!
Let us help you do the math – check our AI dataset project calculator.