Application-Oriented Industrial Edge Computing
摘要
Industrial edge computing applications cover nearly every possible scenario in our daily life. In the current era of AI, the most promising scenario is in the area of edge-assisted model inference, whose typical application is object detection. Object detection is the basis of making any other control decisions and also plays an irreplaceable role in preventive maintenance and quality control. Therefore, this chapter will introduce edge-assisted object detection for two typical data, i.e., image and point cloud. Meanwhile, fluctuations in wireless bandwidth may incur long communication latency for both edge-assisted methods, and thus we propose a teacher–student learning framework to further accelerate the inference.