Optimal Solution for Distance Detection Using Deep Learning Techniques on Embedded Devices
摘要
Deep neural networks (DNNs) are used to solve pattern bases issues like object identification, image classification, annotation, and many others with a high level of accuracy in several fields like manufacturing, medical, autonomous driving, defense and security, etc. These DNN models’ precision comes at the expense of enormous computational overhead, high power consumption, and huge memory consumption. In order to determine the distance between individuals and to represent it effectively in the output, we propose an optimized DNN model using quantization and port it on embedded platform. We validate the same on mentioned dataset and specific device for testing purposes. Our approach includes quantizing a more rapid and precise distance detection model and transferring it to edge devices.