DCAITI Project: Street damage detection and rating

Supervisor: Dr. Witzke, Marcus

Affiliation: Daimler Center for Automotive IT Innovations, Technical University of Berlin, Berlin, Germany.

Overview of the project

1.Labeled and augmented the collected data(GRDDC and ZEB datasets).

2.Trained YOLOv5 Model with transfer learning and Detected damage on ZEB image dataset.

3.Defined scoring model according to the detection results, with labels from shapefile.

• Random Forest Regression, Support Vector Regression, Particle Swarm Optimization(PSO).

See detail on the presentation slide and paper; programming part in branch master.