Deep Vision Interoperability Specification Standard (DeepVISS) aims at providing a common framework for storing and processing the event-driven output of various computer vision workloads, so as to shorten the time-to-market for innovation, to reduce the complexity of integration between independently-developed solution and to enable cross-domain logic between event streams.
Deep Vision Interoperability Specification Standard (DeepVISS) aims to shorten the time to market of computer vision innovation by providing a event-based framework for integration, processing and consumption of outputs from various algorithms, spanning over several domains, such as face detection/identification, car and license plate recognition, pedestrian tracking and many others. Our mission is twofold.
While on the technology level, the DeepVISS framework enables various computer vision workloads to seamlessly connect together and with other systems and to usher the way for higher-level abstractions such as second-order events and practical applications of machine learning. One example would be of a face recognition and classification system and of a automatic license plate recognition (ALPR) system exchanging data and events in order to facilitate correlations between cars and owners.
On the community level, it aims to provide a forum for dialogue between the cutting-edge innovation and state-of-the-art capabilities of the research community and the practical applications envisioned by industry leaders already pursuing a computer vision strategy within established companies and organisations.