Sensors (Basel, Switzerland)
Scrap-SAM-CLIP: Assembling Foundation Models for Typical Shape Recognition in Scrap Classification and Rating.
Guangda Bao, Wenzhi Xia, Haichuan Wang, Zhiyou Liao, Ting Wu, Yun Zhou
Published: 202610.3390/s26020656
Abstract
To address the limitation of 2D methods in inferring absolute scrap dimensions from images, we propose Scrap-SAM-CLIP (SSC), a vision-language model integrating the segment anything model (SAM) and contrastive language-image pre-training in Chinese (…
Preview only. Read the full abstract at the source