DETECTION FOR DAMAGE ON BUILDINGS AFTER EARTHQUAKE USING CONVOLUTIONAL NEURAL NETWORK
Surveying the region is an important step to do for a recovery stage after disaster. By surveying the region, further action to fix the damages can be planned better. In this research, a means to survey the region using satellite imagery and artificial neural network is proposed. The usage of sat...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/80582 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Surveying the region is an important step to do for a recovery stage after disaster. By surveying
the region, further action to fix the damages can be planned better. In this research, a means to
survey the region using satellite imagery and artificial neural network is proposed. The usage of
satellite imagery is common practice for damage assesment after a disaster occured. By adding
an implementation of artificial neural network in the form of Unet-VGG16 model, damage
classification for buildings can be performed. The model has shown a relatively low mistake
indicated by loss function below 0.05. Further tests are carried using image data of the Cianjur
earthquake in November 2022. A qualitative analysis shown that while losing precise details of
the damage, the model has achievedthe purpose of giving accurate general assesment on the
overall damage that occured in the region.
|
|---|