Frost Detection and Thickness Estimation using a Transformer-based Semantic Segmentation Model

doi:10.1016/j.ijrefrig.2026.106899

Frost Detection and Thickness Estimation using a Transformer-based Semantic Segmentation Model

2026-03-09

0

PRE

AI

DOI：10.1016/j.ijrefrig.2026.106899

Original

Original request for help

Save

Abstract

En 中文

• A novel Transformer-based approach for frost detection and quantification • Transformer SegFormer model enables accurate frost detection and estimation • SegFormer-B1 achieved 93% accuracy and 0.875 IoU in frost segmentation • Frost pixel to thickness regression showed strong correlation with R² of 0.93 • Proposed method outperformed thresholding and K-means with 11.98% minimum error

Keywords:

Frost detection

Transformer-based model

Semantic segmentation

Frost thickness estimation

SegFormer

Frost Detection and Thickness Estimation using a Transformer-based Semantic Segmentation Model

Abstract

Journal

Organization