Development Of Object Recognition System By Using Raspberry Pi

Chew, Chiew Yik (2019) Development Of Object Recognition System By Using Raspberry Pi. Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text (24 Pages) Development Of Object Recognition System By Using Raspberry Pi.pdf Download (522kB)

Abstract

This project aims to design and develop an object recognition system that can be used for various purposes such as face recognition, feature recognition and vehicle number recognition. By studying the method used in previous object recognition system, the object recognition system proposed in this project aims to reduce costs, lower power consumption and portable. Single board computer namely Raspberry Pi 3 model B+ is selected as the main processor and Raspberry Pi 8MP camera V2 as the viewing device for this system. Pi camera will capture image or video of an object in the real-time. Then the captured image or video is detected, classified and recognized by using Single Shot Detector (SSD) approach. After the object is detected, the object is bounded by bounding box with the class name of object and the probability. The network of SSD generates scores for object, then the highest score is picked as the class for the bounded object. The accuracy of the system is tested in two different conditions which are good contrast of the environment and plain background. The results show that the accuracy of system in plain background is 90% while the accuracy of system in complex background is 55%. The accuracy of the system in good contrast is 80% which is 45% higher than the accuracy of system in poor contrast.

Actions (login required)

View Item