Side project during my internship at POSCO Assan TST · 2026
What I did
- Converted the RLE masks of a 12,568-image steel surface dataset into YOLO format.
- Trained a YOLOv8n-seg segmentation model on it.
- Presented the results in a Streamlit web app deployed to Streamlit Community Cloud.
Context
My main internship project was finding the process parameters behind surface defects from coil sensor data on the same line — see the experience section on the Technical page.