SGDR vs Slanted Triangular LR, when to use which?

Is Slanted Triangular LR (used in NLP problems in pt2) used in image classification/structured data? Or is SGDR still better for those cases?