DeepLook is a large-scale cloud-based video analytics technology that integrates video streaming and computing resources.
ITRI showcased its game-changing technology for city security surveillance, DeepLook, at CES 2018. DeepLook is a large-scale cloud-based video analytics technology that integrates video streaming and computing resources. It monitors traffic activities so that public safety and law enforcement agencies can respond rapidly to reported incidents.
DeepLook enables license plate recognition, video condensation, and camera network monitoring. It leverages existing video surveillance infrastructure and incorporates cloud computing capabilities to enhance scalability and flexibility of incident analysis, and mass-scale video analysis, condensation, and summarization.
DeepLook is designed with a platform-as-a-service (PaaS) architecture and features a variety of VA/AI/DNN engines, an intuitive plug-and-play user interface, a cloud-native deployment, and a malleable cloud platform. It can integrate in seconds any video or image analysis engines (including the GPU supported AI/DNN model), and it automates and transforms time- and resource-intensive chores into ready information that field officers can use. For instance, an image sharpness detection model can help manage a surveillance network, ensuring all video recordings are clear enough for immediate interpretation. Furthermore, DeepLook can be extended to support various kinds of identification engines and expanded to support many applications.
Potential applications of DeepLook include vehicle tracking, tailgating detection, suspicious activity monitoring, automatic license plate identification, and camera network health status management. To date, DeepLook has been deployed by several regional police stations in Taiwan to support their investigation tasks, for example, tracking stolen cars and those associated with traffic accidents or violations. It has become an effective tool to accelerate the image identification process and thus reduce the work load of the police. It is expected that in the future DeepLook can be adopted when local governments build their video surveillance infrastructure, helping fight crime and improve city safety.