Ground Truth #8

It's getting colder but you can always stay warm with a cup of coffee and a new edition of our newsletter. Scroll down to read more…

From the Industry

In October, the biggest news in the synthetic data space was the release of?DigiFace-1M: 1 Million Digital Face Images for Face Recognition.

?Armed with 1.22 million photorealistic face images with an impressive 110,000 unique identities, DigiFace-1M is the largest public synthetic face dataset to date.

Rendered with a computer graphics pipeline, the dataset is photorealistic and diverse. It consists of synthetic identities which are generated with random facial geometries and textures. The same identity can have different hairstyles and random accessories. To simulate the conditions of real-world data, every identity is then given a variety of poses, facial expressions, lighting conditions, camera angles, and background.

?Scroll down to read our?review of the paper.

From Academia

Paper Recommendation: DigiFace-1M: 1 Million Digital Face Images for Face Recognition

The paper stands on the shoulders of giants; it reused the existing state-of-the-art face generation pipeline (“Fake it Till you Make It” by Wood et al). This pipeline boasts its ability to synthesize face images with minimum domain gap.

DigiFace-1M varied its face images with extensive data augmentation. Real-world images are imperfect – shaky hands lead to blurry photos and poor photography skills cause occluded faces. Instead of only including picture-perfect images, DigiFace-1M adds flaws to its images through a series of flipping, cropping, wrapping blurring, downsampling, and compression to better mimic real-world conditions.

?Read the paper review?here.

From the Team

Customizable Infrared

We have just introduced a new notebook that lets you dynamically tweak the faces infrared renders and fine tune them to meet your exact specifications. Contact us for more information or start generating faces on your own?here.?

Fresh New Content

eBook: How to Use Synthetic Data in 6 Easy Steps?

Once you’ve generated your synthetic dataset, what’s next? Starting with a data-centric model, our new eBook walks you through using synthetic data in your model, testing and training, working with edge cases and controlling bias. Download?here.?