One of the main challenges of implementing AI computer vision applications for mobile systems is reliability and data availability. Even though the on-vehicle video analytics with DeepInsights AI are comprehensive, there will frequently be a request to review archived videos to search for additional information. The weight and power constraints for vehicles will determine the amount of video data storage and, therefore, retention time that can be maintained on the bus before data is overwritten.
The instance of EPIC iO's DeepVision, deployed both on the Bus and at a bus depot, manages the periodic offloading and storage of video data from the fleet of buses that periodically use the depot. The schedule of buses coming and going from the depot is independent of the status of data offloading. The data offload manager has to manage connections to the buses over private 5G and keep track of where to start offloading data based on the status of where offloading stopped during the last connection.