Steel Defect Detection

Python, OpenCV, Streamlit, ONNX

General Info

Because of heating and rolling, drying, and cutting, various equipment contact flat steel throughout the manufacturing process. As a result, steel has been scrapped from the supply chain. After creating a picture of steel, we need to identify any segmentation faults in the material. Segmentation models were built using HarDNet, while the website itself was built using Streamlit.

Output

Below is an example of the detection result on the local web application.

Screenshot of the detection result on the local web application.