Confidence-aware agglomeration classification and segmentation of 2D microscopic food crystal images

TL;DR

This paper introduces a confidence-aware classification and segmentation method for 2D microscopic food crystal images, improving accuracy in identifying agglomerated crystals despite annotation challenges.

Contribution

It presents a novel combined classification-segmentation approach with confidence-aware evaluation, addressing annotation difficulties in microscopic food crystal analysis.

Findings

Enhanced true positive agglomeration classification accuracy

Improved size distribution predictions

Effective classification under varying confidence levels

Abstract

Food crystal agglomeration is a phenomenon occurs during crystallization which traps water between crystals and affects food product quality. Manual annotation of agglomeration in 2D microscopic images is particularly difficult due to the transparency of water bonding and the limited perspective focusing on a single slide of the imaged sample. To address this challenge, we first propose a supervised baseline model to generate segmentation pseudo-labels for the coarsely labeled classification dataset. Next, an instance classification model that simultaneously performs pixel-wise segmentation is trained. Both models are used in the inference stage to combine their respective strengths in classification and segmentation. To preserve crystal properties, a post processing module is designed and included to both steps. Our method improves true positive agglomeration classification accuracy and size distribution predictions compared to other existing methods. Given the variability in confidence levels of manual annotations, our proposed method is evaluated under two confidence levels and successfully classifies potential agglomerated instances.