Open AccessReal-time and accurate object detection on edge device with TensorFlow Lite
Author(s) -
Jiyan Dai
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1651/1/012114
Subject(s) - computer science , artificial intelligence , edge device , enhanced data rates for gsm evolution , inference , object (grammar) , computer vision , object detection , robotics , deep learning , artificial neural network , window (computing) , pattern recognition (psychology) , robot , cloud computing , operating system
Objection detection is of vital importance to many fields, such as autonomous driving, outdoor robotics, and computer vision. Existing approaches on object detection can hardly run on the resource-constrained edge devices. In order to mitigate this dilemma, we propose to apply TensorFlow Lite to convert Float32 neural network model to unit8 neural network with subtle or even no accuracy loss. Two advantages are here for conversion. First, it reduces the model size to a quarter so that it fits for devices with limited storage. Second, it achieves much faster inference time. I conduct an experiment on MSCOCO dataset. Experimental results show that our proposed method achieves mAP 72.1 and FPS 23 on edge device.