of Concrete Structures
Based on drone images of concrete surfaces, structural defects are detected and classified (assigned to a specific defect type, e.g. cracks, spallings, reinforcements).
Artificial intelligence is used to handle the huge variety of appearances within a specific defect type. It also ensures that uncritical concrete structures, such as joints or rough surfaces, which almost look identical to real defects, are not misclassified.
The use of drones and the huge variety of the image recording equipment leads to extremely fluctuating image properties (brightness, distance to the concrete structure, sharpness etc.). The software is able to handle these fluctuations, thus ensuring very reliable results.
extremely fluctuating quality of original images
classification of very small defects: 0,1 mm
Original Image: crack
Result Image: crack
Original Image: crack + spalling
Result Image: crack (orange) + spalling (pink)
Original Image: spalling + reinforcements
Result Image: spalling (pink) + reinforcements (blue)