Automotive Photography
Improving the quality of car photography using Computer Vision
Will Perkins
Business Challenge
The emergence of online automotive marketplaces has revolutionized the way consumers buy and sell pre-owned vehicles.
Quality photography is crucial for car buyers to confidently reach a decision about purchasing a pre-owned vehicle. Car sellers struggle to capture high-quality photos due to their limited knowledge of photography and lack of access to professional equipment.
“ With the help of Computer Vision, we can guide a car seller to take better photos.”
Approach
Laan Labs approached this problem for a German automaker by creating a mobile app that enables a car seller to quickly and easily photograph the exterior and interior of a car using an ordinary smartphone. With the help of Computer Vision, we can use software algorithms to guide a car seller to take better photos.
Powered by a custom-trained neural network, the app estimates the pose of a car and provides visual feedback to the user in real-time: helping to position a car in the camera's frame - to obtain the perfect shot every time.
Some of the tasks performed with Computer Vision include:
-
Car Pose Estimation
-
Sufficient Lighting Detection
-
Automatic Camera Focus
Business Value
By leveraging Machine Learning and state-of-the-art Computer Vision techniques, Laan Labs was able to quickly deliver an iOS app, with a simple & intuitive experience that produces high-quality photos with strong visual consistency.
Technologies Utilized
Computer Vision, Neural Networks, Augmented Reality, Python, iOS, TensorFlow, CoreML