This paper proposes a surveillance monitoring system for wild animal named W-MERCI. The monitoring system consists of two Raspberry Pi as surveillance nodes which integrated with detectors and actuators node. Detectors comprises of NoIR camera and Ultrasonic sensor, while actuators are mini speaker and LED. The proposed system is to help farmer to protect their farm from the wild animals. In the NoIR camera, we also deploy a combination od MobileNet and Single Shot Detectors (SSD) for object detection using deep neural network (DNN). The combination of both methods produces a fast and real-time object detection on resource constrained devices. From the DNN process, an optimization is performed which produce a higher output frame-per-second (FPS) up to 5 FPS. To power the surveillance nodes, this paper uses solar energy harvesting which can run the system for approximately two hours per day.
|Attachment 1:||MERCI for Wild Animal Final Presentation.pdf(2.0MB)|
|1||[2019 Spring Semester] W-MERCI: Wild Animal Surveillan..||°ü¸®ÀÚ||2019-07-15||6936|