Road Accident Analysis for Community Awareness
Team
- Pubudu Bandara (E/17/027) - email, GitHub
- Dhananjaya Morais (E/17/212) - email, GitHub
- Ishara Nawarathna (E/17/219) - email, GitHub
Table of Contents
- Introduction
- Problem
- Motivations
- Objectives
- Requirements
- Solution
- Solution Architecture
- High Level System Organization
- Data Flow of the System
- Technologies used
- Use Case diagram
- Machine Learning Model
- Process
- Workplan
- Future work
- Links
Introduction
Road traffic injuries are a leading cause of death and disability. In low and middle income countries, vulnerable road users are commonly involved in crashes with severe injuries. Road traffic injuries are a major public health problem globally. About 1.2 million people are killed and more than 50 million are injured due to road traffic crashes annually.More than 90% of these deaths and injuries occur in low and middle income countries (LMIC) due to rapid motorization, lack of road safety culture, poor road conditions, and lack of education on road safety. Our aim is to reduce road accidents using machine learning techniques.
Problem
The prevailing strategies are not sufficient enough to reduce the frequency and severity of road accidents.
Motivations
- The Daily death toll from road accidents
- Inefficiency of police traffic management
- Difficulty in handling the casualties in hospitals.
Objectives
- Reduce Road Accidents
- To identify the main factors associated with a road accident (accident data analysis).
Requirements
Functional Requirements
- Data Visualization
- Obtain Main Factors of the Road Accidents
- Ability to upload new Datasets
- Predict the future Accidents
Non-Functional Requirements
- Performance of the Prediction Model
- Improved UI / UX
- Security
- Scalability
Solution
An online system to provide more accurate information on road accidents (both analysis and predictions). We are going to build a real time web application which can be used by the public without logging to the system.
Solution Architecture
High Level System Organization
Data Flow of the System
Technologies used
- Motor: Asynchronous Python driver for MongoDB
- FatsAPI: Web framework for developing RESTful APIs in Python
- MongoDB Atlas: Cloud database service
- Asyncio: Library to write concurrent code, often a perfect fit for IO-bound and high-level structured network code
- React: JS library for building UIs
- Pandas: Data analysis and manipulation tool
- TensorFlow: Software library for ML and AI
- Keras: Software library that provides a Python interface for ANNs
Use Case diagram
Machine Learning Model
Workflow of the Machine Learning Process
- Data Collection - US Accidents Dataset(2016 - 2021) from Kaggle
- Data Preprocessing
- Develop the machine learning model
- Model Fitting
- Model Evaluation
- Model Deployment
Process
Workplan
Future work
Robust Data Handling
- Datasets with different attributes
- Resulting in a more accurate and up-to-date mode
Realtime Modeling
- Users can analyze and predict on their own datasets
Current Progress
- Analyzing the Dataset
- Create API endpoint
- Develop Homepage
