Car Damage Detection with Deep Learning Algorithms

about
Our client is an international car rental company.  Upon return, the vehicles are manually inspected for damages by experts.
Problem
  • Damage control and related cost calculations are being performed by experts.
  • Due to the increase of fleet and inspection quantity, alternative solutions are required for cost and time optimization
Action
  • Developed an AI algorithm to replace experts and their manual controls.
  • Combined two models (1. part detection 2. damage detection) to predict required replacements and repairs.
  • Integrated with third party to obtain up to date spare part costs
Tool Stack
  • Azure, Python, Rabbit MQ, .NET, SQL DB, Vue.js, Pytorch, Tensorflow

95

%

Accuracy rate for detection

400

%

Increase process cycle

85

%

Automation