This project aims to classify the status of blast furnace into 5 categories
(0: normal, 1: Non-blowing of pulverized coal, 2: Poor combustion, 3: Iron-ore falling, 4: Coke Turning)
The classification model showed above 90% accuracy.
For introducing the model on industrial floor,
Grad-cam is applied in order to explain reasoning area of blast furnace images.
The result example is below.
The data and trained model can't be open for security reasons.
However, I opened used source code of simple classification model and grad-cam.