AI BriefingML Use Case for Turkey Earthquake Rescue
Key point
This covers a case where machine learning was used during the Turkey earthquake to structure rescue request information from social media and support rescue activities.
Details
Immediately after the Turkey earthquake in February 2023, the afetharita (disaster map) application was developed to process rescue request information posted on social media in real time.
The key technical implementation details are as follows:
- Data extraction: Text within images was extracted via
easyocr, and a Transformer NER model fine-tuned frombert-base-turkish-casedwas used to structure key information such as names, phone numbers, and addresses. - Interface and deployment: An interface was built with
Gradioand hosted on Hugging Face Spaces, providing stable service through GPU support. - Model serving: The Hugging Face Inference API was used to deploy the model immediately without building separate infrastructure, securing scalability to handle traffic surges.
The extracted addresses were converted into coordinates via the Geocoding API and displayed on the map, which played a decisive role in helping rescue workers quickly locate survivors.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.