At the F8 2018 Day 2 in San Jose, California yesterday, May 3, Facebook provided details on three areas it is focusing on long-term investments. These are connectivity, Artificial intelligence, and AR/VR.
While going forward with advances on AI and AR/VR, a ‘strong internet connection’ is the prime necessity, and about 3.8 billion people in the world do not have access to Internet as Facebook says. The company is developing next-gen technologies to reduce cost and increase performance and capacity for a larger mass to have access.
Regarding connectivity works of Facebook, it partnered with local operators in Uganda in bringing new fiber and providing backhaul connectivity when completed. This will cover over 3 million people, and as well will enable future cross-border connectivity to neighboring countries.
Additionally, the company’s team in San Jose started testing an advanced Wi-Fi network that’s supported by Terragraph. Similar trials will be carried out in Malaysia and Hungary. Facebook further has collaborations with ‘hundreds of partners’ belonging to the Telecom Infra Project that aims at creating and launching various innovative and efficient network infrastructure solutions. Lastly, the social media giant is sharing its experiences on connectivity for other relevant service providers to benefit from.
As for AI, Facebook sees it as a foundational technology. One of its recent breakthroughs is its successful training on an image recognition system that includes a data set of 3.5 billion publicly available photos.
The system is powered by Facebook’s AI research and production tools: ONNX, Caffe2, and PyTorch. Yesterday, Facebook announced its open source AI framework’s next version, the PyTorch 1.0. As described, the framework “combines the capabilities of all these tools to provide everyone in the AI research community with a fast, seamless path for building a broad range of AI projects.”
PyTorch 1.0 toolkit will have its beta version available within the coming few months. Developers will find it useful in benefitting from computer vision advances such as DensePose for making AR camera apps more immersive.
For the area of AR/VR, Facebook’s research scientists created a prototype system to generate 3D reconstructions of physical spaces. Experiencing ‘surprisingly convincing results’, the company shared a video that compares 3D reconstruction and normal footage. The demonstration screams out the real-like capabilities of the system! Facebook is further working on state-of-the-art AR/VR research for computers to come up with photorealistic avatars.
Keep following us to get more info on innovations of Facebook.