World Health organization reports every year, we lose approximately 1.35 million people due to road accidents. Between 20 and 50 million more people suffer non-fatal injuries, with many incurring a disability as a result of their injury. In Kenya, over 3000 people die every year due to road accidents and approximately 10,000 suffer injuries.
What it does
Quick response to assist victims of road accidents is important to reduce the severity of injuries. AccidentRecognizer classifies accident images to be able to detect and recognize occurrence of an accident using video footage from traffic cameras. AccidentRecognizer sends an alert notification to emergency response service provider notifying them to confirm occurrence of an accident and act to rescue the victims.
How I built it
- I build a Convolutional Neural Networks to classify images using Python and PyTorch machine learning library.
Challenges I ran into
Shortage of time to collect data for training and testing neural network
Accomplishments that I'm proud of
- Development of AccidentRecognizer to help save lives
What I learned
- Different types of Neural Network and Applications
- Optimization of neural networks
- Training of neural network
What's next for AccidentRecognizer
- Completion of the project
- Integration with traffic cameras.
- Partnership with actors providing emergency response and rescue services.
- Collaboration with public and private entities in transport and health sector.
- Development of artificial neural network to predict occurrence of an accident and alert motorists to help save lives.