Social Items

Facebook's experiments on both AR and VR has reached an advance stage, both for entertainment and communication purposes, with the company now working on AR applications that could not only modify or replace the face, but the entire body.

The company's AI Camera Team which has been credited for various computer vision technologies and creative tools that help people express themselves, has come up with real-time “style transfer” - Mask R-CNN, as the technique is called, to give your photos or videos the look of something like a Van Gogh painting.



While the research is far from unprecedented; as skeletal tracking systems are commonplace in many industries, Facebook is looking especially into keeping things simple and easy to deploy on mobile.

The aim is to ease working within hard limitations as relate to image resolution and sensor data, with the processing power available.

Albeit, the development of computer vision models for mobile devices remains a challenging task, as a mobile model doesn't only require to be small, but fast and accurate without large memory requirements.

The company has promised to continue tinkering with different models that could better fit into mobile GPUs and DSPs which has the potential to save both the battery and computational power.

Facebook tinkering with full-body replacement/tracking in a VR or AR context

Facebook's experiments on both AR and VR has reached an advance stage, both for entertainment and communication purposes, with the company now working on AR applications that could not only modify or replace the face, but the entire body.

The company's AI Camera Team which has been credited for various computer vision technologies and creative tools that help people express themselves, has come up with real-time “style transfer” - Mask R-CNN, as the technique is called, to give your photos or videos the look of something like a Van Gogh painting.



While the research is far from unprecedented; as skeletal tracking systems are commonplace in many industries, Facebook is looking especially into keeping things simple and easy to deploy on mobile.

The aim is to ease working within hard limitations as relate to image resolution and sensor data, with the processing power available.

Albeit, the development of computer vision models for mobile devices remains a challenging task, as a mobile model doesn't only require to be small, but fast and accurate without large memory requirements.

The company has promised to continue tinkering with different models that could better fit into mobile GPUs and DSPs which has the potential to save both the battery and computational power.
Emerging Technology

No comments

Subscribe to: Post Comments (Atom)