Abstract:Spectroscopy-based techniques have made remarkable advancements in their application to fruit quality control but encounter challenges of high model uncertainty arising from biological variability. Minor changes in spectral measurement orientations or positions significantly altered the model prediction for fruit quality, which severely affects the reliability of online fruit grading systems. This study assesses the influence of varying structural designs in deep learning models and develops a Knowledge-Guided Convolutional Neural Network (KGCNN) to mitigate predictive uncertainty caused by different orientations. A novel loss criterion that quantifies model uncertainty incorporating a grouped input strategy is introduced, enabling the implementation of a knowledgeguided approach. Experiments are conducted on Mandarin and Valencia orange datasets collected on an onsite grading system for soluble solid content assessment. The proposed KGCNN model effectively reduces 27.1 % and 25.5 % averaged prediction variance compared with CNN and PLS models on five detections of random orientations, demonstrating a significant mitigation of model uncertainty. This strategy operates within the framework of the developed model without the need for structural hyperparameter adjustments, with a minor impact on the test accuracy, evidenced by a 1.7 % increase in RMSE and a 0.2 % decrease in R2 averaged on two datasets. A hidden feature visualization technique is further employed to explain the working mechanism of the decreased model uncertainty by KGCNN. The visual evidence supports the effectiveness of the knowledge-guided approach in mitigating the nonlinear scatter within the distributions of learned features from input data collected at multiple orientations, attributed to the constraint effect of the introduced loss criterion. This research has the potential to alleviate the stringent requirements on fruit loading or measurement positions within onsite grading systems and enhance the reliability of deep learning models in practical applications.

a deep learning-based framework for secondary metabolite class prediction using fruit and leaf images

Fruit-In-Sight: A deep learning-based framework for secondary metabolite class prediction using fruit and leaf images

Spec2Class: Accurate Prediction of Plant Secondary Metabolite Class using Deep Learning

Fruit and vegetable leaf disease recognition based on a novel custom convolutional neural network and shallow classifier

Fruit classification using attention-based MobileNetV2 for industrial applications

Fruits Classification and Detection Application Using Deep Learning

Deep Learning-Based Fruit Recognition and Quality Assessment: A Convolutional Neural Network Approach

Fruit Classification System with Deep Learning and Neural Architecture Search

An Application of Deep Learning for Sweet Cherry Phenotyping using YOLO Object Detection

Prediction of Fruit Maturity, Quality, and Its Life Using Deep Learning Algorithms

A convolutional neural network-driven computer vision system toward identification of species and maturity stage of medicinal leaves: case studies with Neem, Tulsi and Kalmegh leaves

Plant-Leaf Disease Prediction Using Deep Learning

Deep neural network based fruit identification and grading system for precision agriculture

Portable near-infrared spectral imaging combining deep learning and chemometrics for dry matter and soluble solids prediction in intact Kiwifruit

Robust deep learning method for fruit decay detection and plant identification: enhancing food security and quality control

Enhancing Spectroscopy-Based Fruit Quality Control: A Knowledge-Guided Machine Learning Approach to Reduce Model Uncertainty

A novel framework for semi-automated system for grape leaf disease detection

Assessing deep convolutional neural network models and their comparative performance for automated medicinal plant identification from leaf images

DeepFruits: A Fruit Detection System Using Deep Neural Networks

Facilitated machine learning for image-based fruit quality assessment

Citrus leaf disease detection through deep learning approach