Labelability of extremely low-resolution manufacturing images

Determine whether in-control and out-of-control manufacturing images containing only a few pixels can still be distinguished reliably and labeled correctly for image-based quality monitoring.

Background

The paper shows that the convergence rates of MODERN-Net and MODERN-Diagnosis deteriorate as image dimension, or resolution, increases, motivating the possibility that manufacturers could use lower-resolution images. However, reducing resolution may make it difficult to distinguish in-control images from out-of-control images and to assign correct labels. The authors explicitly raise this question before analyzing the effect of label misclassification, but the paper does not establish whether images with only a few pixels remain reliably classifiable.

References

We note, however, an important question arises in such a scenario concerning whether we can still tell the OC images apart from those IC images, with each image having only a few pixels. That is, are we still able to label the images correctly?

— Deep Vision in Smart Manufacturing: MODERN Framework for Intelligent Quality Monitoring and Diagnosis  (2608.13937 - Kang et al., 14 Aug 2026) in Subsection 4.1, subsection “Practical implications - part 1”