Real-Time Object Detection with Intel NCS2 on Hardware with Limited Resources for Low-power IoT Devices

This paper presents several models for real-time object detection with a hardware extension on hardware with limited resources. Additionally, a comparison of two approaches for detecting individual objects with Single-Shot Multibox Detection (SSD) and You Only Look Once (YOLO) architecture in a 2D image with Convolution Neural Networks (ConvNet) is presented. Here, we focus on an approach to develop real-time object detection for hardware with limited resources in the field of the Internet of Things (IoT). Also, our selected models are trained and evaluated with real data from model making area. In the beginning, related work of this paper is discussed. As well known, a large amount of annotated training data for supervised learning of ConvNet is required. The data acquisition of the different real data sets is also discussed in this paper. Additionally, our dissimilar object detection models are compared in accuracy and run time to find the better and faster system for object detection on hardware with limited resources for low-power IoT devices. Through the experiments described in this paper, the comparison of the run time depending on different hardware is presented. Furthermore, the use of a hardware extension is analysed in this paper. For this purpose, we use the Intel Neural Compute Stick 2 (NCS2) to develop real-time object detection on hardware with limited resources. Finally, future research and work in this area are discussed.