Machine learning for object detection

Abstract Artificial Intelligence (AI) and its applications have continuously gained traction in recent years, with visual understanding as part of Computer Vision (CV) representing one particularly demanding application. Especially for real-time scenarios, the stringent QoS requirements combined with high bandwidth needs for image or video data represent a particular challenge. In this chapter, we describe the implementation and evaluation of a distributed object recognition service within Service Function Chainings (SFCs), which can be optimal for deploying object detection services, especially when facing simultaneous network traffic and low latency requirement challenges. In the hands-on examples of this chapter an optimized distributed AI approach to real-time visual understanding will be evaluated within the ComNetsEmu environment.