Apple Unveils AI Research Unleashing 3D Image Depth Mapping Using 2D Input Images

The Apple research team has unveiled their Depth Pro AI model, which can generate three-dimensional depth maps from two-dimensional image inputs in less than a second, without relying on additional camera data. Operating by predicting image depth through training on high-resolution image datasets, the Depth Pro model can differentiate between object edges and provide depth levels for each position in the image.

A standout feature of Depth Pro is its speed, capable of generating high-resolution 3D image maps of 2.25 megapixels in just 0.3 seconds using only one standard GPU. This makes the model applicable for tasks requiring image depth assessment, such as AR or even self-driving cars.

For more detailed information on this research, visit GitHub.

TLDR: Apple’s research team introduces the Depth Pro AI model for quick and accurate 3D depth mapping from 2D images, suitable for various applications including AR and autonomous vehicles.