Biology for Computer Engineers Course Handout.pptx
INVESTIGATION OF CRACKS ON BUILDINGS.pptx
1. USING COMPUTER VISION AND DEEP
LEARNING
INVESTIGATION OF CRACKS ON
BUILDINGS
SIHAAM S
III rd YEAR (2020-2024)
DEPARTMENT OF ARTIFICIAL
INTELLIGENCE AND DATA
SCIENCE `
2. ABSTRACT
Investigation of cracks such as concrete
cracks, steel frame cracks and damages on
huge buildings like apartments, bridges, dams
etc using computer vision –image processing
and deep learning for which manual
inspection consumes time and cost
Image-based crack detection method using a
deep convolutional neural network (DCNN).
3. CRACKS ON BUILDINGS
Cracks on buildings occur both naturally and by
human error.
After few years once a building is constructed proper
inspection is needed to check the cracks and
damages in the building and immediate repair and
reconstruction is required to make the building
resilient
Cracks on buildings reduces the bearing capacity,
durability and waterproofing of concrete structures.
If the cracks are left unchecked and unrepaired it
increases the probablity of the building collapsion
5. DEVELOPING THE DCNN MODEL
COLLECTING THE DATASET – Images of various kinds of cracks and
damages
DATA AUGMENTATION
Good quality and quantity data is required to build the model with high
accuracy
6. METHODOLOGY
CREATING THE DEEP CONVOLUTIONAL
NEURAL NETWORK layers such as INPUT
LAYER, HIDDEN LAYERS , OUTPUT LAYER
7. Model working
Input image 1st layer 2nd layer nth layer output
For example
Cracked image 20% crack detected 45% crack
detected 90% crack detected image classified as
cracked image
9. Conclusion
As a result, an autonomous and
intelligent approach for collecting
damage/crack information from images
or videos is required to reduce human
labor and aid servicing engineers in
speeding up the inspection process
while maintaining accuracy.
10. Artificial Intelligence is really powerful when
it integrates with construction and
machineries AI can create wonders