Paper - A Comparative Analysis of Computer Vision Libraries in the Context of a Jakarta Traffic Simulator

Author: Ida Bagus Kerthyayana Manuaba, Thomas Dwi Dinata

Traffic management is essential for preventing traffic congestion and regulating traffic flow. Jakarta has implemented several methods to reduce traffic congestion in response to the growing number of vehicles on the road. This study proposes integrating computer vision and fuzzy logic into traffic lights for use in traffic management. The study focuses on comparing and implementing three computer vision libraries: ImageAI, Cloud Vision, and OpenCV Libraries. This comparison is based on Jakarta traffic data captured by closedcircuit television (CCTV). After evaluating the libraries, the result indicates the ImageAI achieves better performance than the other libraries across testing measurements and it was used to develop a module simulator. This simulator was used to compare the amber time produced by each lane at a simple intersection under fixed time versus proposed-actuated traffic lights. The simulation of the proposed actuated traffic light scenario demonstrates that the use of computer vision and fuzzy logic in the decision-making process can marginally improve traffic simulation compared to the fixed-time traffic light scenario. This study illustrates the application of computer vision and fuzzy logic to traffic management. The application of these technologies to traffic lights has the potential to increase traffic flow and decrease congestion.

Thesis - MONAS: Intelligent Traffic Light System, Implementation of Computer Vision and Fuzzy Logic to Traffic Light

Traffic management is an important aspect on the road to prevent traffic jams and regulate the traffic. In Jakarta, several techniques have been implemented to reduce the traffic congestion caused by the increasing vehicles inside Jakarta. The author proposed the implementation of computer vision and fuzzy logic to be implemented into the traffic management by embedding it to the traffic light. The experiment will be focusing on comparing three computer vision libraries and developing a simulator used to differentiate fixed-time traffic light with actuated traffic light. The chosen computer vision library that is suitable for vehicle detections in Jakarta later will be used for the simulator as a module. The simulator itself used to proof that using actuated traffic light may able to improve the traffic management.

Cricket Butler

A project to transform unique robotic educational board into a co-processor for other embedded boards to take advantage of it's features such as built-in motor driver, several sensor input and digital input.

Git URL: https://gitlab.com/thomasdwidinata/cricket-butler

Klippy

A project to host file server while keeping in mind of authentication both server and user side, making it trustless authentication that always require authentication even if the incoming request is from same or "trusted" network. It was built using Node.js Express and built in API like style.

Git URL: https://gitlab.com/thomasdwidinata/klippy

Auqey

A sample Typescript project which will be used for my project template and also for anyone who wants to learn Typescript in a fashionable way. This project is built in mind of to be an authentication server and API web server.

Git URL: https://gitlab.com/thomasdwidinata/auqey

Telebot

A quick template for creating Telegram bot complete with routing and automatic scanning of new route files. Written in Typescript!

Git URL: https://gitlab.com/thomasdwidinata/telebot/

EnviroSense

A project from RMIT University Programming Internet of Things (COSC2674), where the aim of this project is to create a environment sensor using Raspberry Pi. The aspect of developing this project ranges from learning to use GPIO and accessing it using Python, using Raspberry Pi's OS which is based on Linux and how to deploy the app on it.

Git URL: https://gitlab.com/thomasdwidinata/iot-envirosense

Huffman Compression

A C++ project which aims to implement huffman coding algorithm to compress a file, which will scan file binary and determine which bits should be shortened using hashmaps and huffman algorithm.

Git URL: https://gitlab.com/thomasdwidinata/file-compression-huffman