Elevator Pitch
Your first salary just got credited and you’re excited, you wanna gift yourself a car. You go to the dealer’s shop. You wanna ensure you don’t get a car that perhaps has some dents or scratches. Imagine you have an app that can detect that with high accuracy. You go home with your car, happy.
Description
In recent years, Computer Vision has had a very rapid development, from facial image classification to self-driving cars, with Convolutional Neural Networks still advancing. Automated Car Damage Detection is one of several applications of CNNs. This can help save time by cutting down the need for manual assessment. Car damage detection will make use of transfer learning of suitable pre-trained CNNs models. A basic overview of how the model works are: 1) Extracting area of damage 2) Classifying if there’s a damage or not
To get the project started, the following steps are taken: 1) Data Collection- collect photos of cars from the internet 2) Data Annotation- map out the areas of damage from the dataset 3) Loading datasets and Training the Data 4) Model validation and Prediction
This technology can be used in the car industry, to assess the state of a car automatically, faster.
Notes
The technical requirements for this project are: 1) Understanding of the Python language 3) Understanding of Machine/Deep Learning 2) As the model is a large one, a GPU PC is required or the use of cloud platforms like Google Colab, Kaggle Kernels, etc.
Ever since I got into tech, I’ve had fulfillment giving back to the community the little knowledge I have. This is one of the reasons I got invited to take a session at GDG Ogbomoso. I had my first share of PyCon experience at the 2018 edition, where I was privileged to serve as a volunteer. I might not necessarily be the best to give this particular talk, but the thought that I have a chance to give back what I’ve learned to a large audience makes me passionate about giving a talk.