Interpretable Citrus Fruit Quality Assessment Using Vision Transformers and Lightweight Large Language Models
Open Access
- 3 September 2025
- journal article
- research article
- Published by MDPI AG in Agriengineering
- Vol. 7 (9) , 286
Abstract
This study introduces an interpretable deep learning pipeline for citrus fruit quality classification using the ViT-Base model vit_base_patch16_224 and Microsoft’s Phi-3-mini LLM. The ViT model, fine-tuned on resized 224 × 224 images with ImageNet weights, classifies fruits into good, damaged, and rotten categories, achieving 98.29% accuracy. For interpretability, Grad-CAM highlights damaged regions, while the Phi-3-mini generates human-readable diagnostic reports. The system runs efficiently on edge devices, enabling real-time, on-site quality assessment. This approach enhances transparency and decision-making, showing strong potential for deployment in the citrus industry.Keywords
This publication has 38 references indexed in Scilit:
- Hyperspectral Imaging and Deep Learning for Quality and Safety Inspection of Fruits and Vegetables: A ReviewJournal of Agricultural and Food Chemistry, 2025
- Qualitative and quantitative analysis of Nanfeng mandarin quality based on hyperspectral imaging and deep learningFood Control, 2024
- A Survey on Evaluation of Large Language ModelsACM Transactions on Intelligent Systems and Technology, 2024
- Automatic Detection of Citrus Fruit and Leaves Diseases Using Deep Neural Network ModelIEEE Access, 2021
- A Survey of Convolutional Neural Networks: Analysis, Applications, and ProspectsIEEE Transactions on Neural Networks and Learning Systems, 2021
- An adaptive segmentation method combining MSRCR and mean shift algorithm with K-means correction of green apples in natural environmentInformation Processing in Agriculture, 2019
- The Implementation of K-Means Algorithm as Image Segmenting Method in Identifying the Citrus Leaves DiseaseIOP Conference Series: Earth and Environmental Science, 2019
- Symptom based automated detection of citrus diseases using color histogram and textural descriptorsComputers and Electronics in Agriculture, 2017
- Unsupervised classification methods in food sciences: discussion and outlookJournal of the Science of Food and Agriculture, 2008
- Brightness-invariant image segmentation for on-line fruit defect detectionOptical Engineering, 1998