In an ever advancing world where everybody is aiming to achieve something bigger and better, Facebook AI Research (FAIR) with the similar objective, declared the open-sourcing of Detectron, company's state-of-the-art platform that works towards object detection research. It is a deep learning Framework that works on computer vision object detection. Computer Vision majorly deals with a field that works on how the process of achieving high-level understanding from digital images and video can be made automated.

The project Detectron was conceived in July 2016 with the intent of developing a fast and flexible object detection built on Caffe2, a deep learning framework written in C++, with a Python interface. It was made to expedite with the prospect of computer vision research and has the ability to make augmented reality truly productive. The project has been implemented and supports a variety of Facebook's application including Non-local Neural Networks, Focal Loss for Dense Object Detection and Mask R-CNN. The crucial tasks in real-time object detection are to identify and interpret what system sees. The company has out-sourced the project with a vision to accelerate the development.

The algorithms implemented by Detectron help intuitive models for important computer vision tasks, such as instance segmentation, and have a pivotal role in the unprecedented advancement of visual perception systems that the community has achieved in past few years. Apart from a research point of view, Facebook teams have been holding up this platform to train custom models for a range of different applications with AR and community integrity. After the models are trained they can be deployed in the cloud and on mobile devices.

AR is present at the core level for interacting with the world without being explicitly pre-programmed for a particular environment which needs a cursory understanding of our proximity. For an instance, if you're wearing AR glasses and want to be able to project the oven temperature above the oven, one of the basic needs to get started with the work is to know the wide variety of ovens available along with their look and the places in which they reside. This is one of the most critical tasks that real-time object detection deals with.

The Detectron is easily available under the Apache 2.0 license at GitHub. The company promises to release more than 70 pre-trained models that are extensive performance baselines which are also available to download from its model zoo on GitHub. Still, there are speculations about which teams will use Detectron for AR but one the most obvious guesses is Oculus. By outsourcing the project, the company hopes to achieve the motive of broadening and enhancing the project's scope.