Deep learning enabled Web AR Application

Web Augmented Reality (AR) refers to augmented reality (AR) experiences that are accessed through a web browser rather than an app. Most of the Web AR solutions will encompass object detection, recognition and tracking, which provide intelligent context-aware ability and accurate vision-based tracking ability. The next step in the Web AR pipeline involves machine learning and deep learning, where is more advanced applicable. However, deep learning relies on heavy in terms of computation deep neural networks such as convolutional neural networks, which is not easy to accommodate to the existing web browsers. DeepWAR will provide low level deep learning components implementation to the SolAR platform using model compression and inference acceleration. DevOps methodology will be used for the proposed deep learning pipeline, where the efficient deep learning implementation within the future AR enabled web-browsing solutions will be provided. DeepWAR will be developed further and made it available into the market from Cogninn in 2021 using the “bounty driven development” business model.