In this paper, the authors conducted a comprehensive benchmarking experimental study that compares the inference performance of Face Detection and Faces Recognition on 4 Edge platforms (all Jetson boards) and 5 cloud GPUs, with a total of 294 experiments, generating +600000 data records. They have also considered TFLite and TensorRT optimization frameworks. They demonstrate that the TensorRT optimization provides the fastest execution on all cloud and edge devices, at the expense of significantly larger energy consumption (up to +40% and +35% for edge and cloud devices respectively, compared to Tensorflow). TFLite is the most efficient framework in terms of memory and power consumption while providing significantly less (-4% to -62%) processing acceleration than TensorRT.
The open-source results and interactive dashboards are available here.