Smart Street Light towards Energy Saving and Neighborhood Security
- Resource Type
- Conference
- Authors
- XU, Yuan; TANG, Tongsheng; ZHENG, Yanghao; YANG, Kezhong; XIAO, Zhiyong
- Source
- 2019 IEEE Green Energy and Smart Systems Conference (IGESSC) Green Energy and Smart Systems Conference (IGESSC), 2019 IEEE. :1-4 Nov, 2019
- Subject
- Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
Engineered Materials, Dielectrics and Plasmas
Fields, Waves and Electromagnetics
Nuclear Engineering
Photonics and Electrooptics
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Transportation
Smart Lighting
Human Detection
CNN
HOG
- Language
- ISSN
- 2640-0138
Both street lights and surveillance cameras are very important to improve security at night in public places. These two types of facilities are often installed together but working independently with different purposes. With edge computing power, surveillance cameras can also be the eyes of street lights and work smartly to save electrical energy. This work presents a smart street light system that is able to detect pedestrians using surveillance cameras and adjust brightness intelligently. A new algorithm HDCV was developed in order to run human detection offline with limited edge computing resources. The HDCV algorithm employs histogram of oriented gradient (HOG) features to do fast detection and convolution network (CNN) based methods to do verification. The edge computing system was implemented with Raspberry Pi (RPi). The prototype demonstrated the capability to detect pedestrians accurately and resist interferences from animals. The system is estimated to save more than 30% electrical energy.