Steel product number recognition framework using semantic mask-conditioned diffusion model with limited data

Hyeyeon Choi, Jong Pil Yun, Bum Jun Kim, Hyeonah Jang, WooSang Shin, Sang Woo Kim

Abstract: Highlights•The problem of recognizing product numbers with limited data is addressed.•Product number image containing user defined characteristics can be generated.•Proposed SMDM generates out-of-distribution samples.•Generated samples diversify the data distribution and improve recognition accuracy.