Reproduce “Facilitated machine learning for image-based fruit quality assessment” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Manuel Knott, Fernando Perez-Cruz, Thijs Defraeye
The authors evaluate a frozen DINO Vision Transformer for fruit classification by extracting image feature vectors and training a shallow classifier on local labels. On CASC IFW apples, DINO ViT-B/8 features with an SVC exceed 90% test accuracy with 500 training samples, while the CNN comparisons need at least 1,500. This low-sample advantage was not found on the banana dataset.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.