Recognition of Mould Colony on Unhulled Paddy Based on Computer Vision using Conventional Machine-learning and Deep...

Area of Science:

• Agricultural Science
• Computer Science
• Food Science

Background:

• Mould contamination in unhulled paddy poses significant risks to grain quality and safety.
• Accurate identification and distribution mapping of mould species are crucial for effective management strategies.
• Conventional methods for mould detection can be time-consuming and labor-intensive.

Purpose of the Study:

• To evaluate the efficacy of conventional and deep learning techniques for recognizing mould species and their distribution in unhulled paddy.
• To compare the performance of Support Vector Machine (SVM), Back-Propagation Neural Network (BPNN), Convolutional Neural Network (CNN), and Deep Belief Network (DBN) models.
• To develop an efficient and accurate method for automated mould detection in unhulled paddy.

Main Methods:

• Unhulled paddy samples were inoculated with five mould species and images were captured.
• Mould recognition models were developed using SVM, BPNN, CNN, and DBN.
• Image processing techniques, including pitch segmentation, were employed for colony area recognition.

Main Results:

• The Deep Belief Network (DBN) model achieved 100% accuracy in identifying mould species using colour-histogram parameters.
• Support Vector Machine (SVM) and Convolutional Neural Network (CNN) models showed approximately 90% accuracy in pitch classification.
• CNN and DBN models exhibited faster calculation speeds for mould recognition.

Conclusions:

• Deep learning models, particularly DBN and CNN, show significant potential for accurate and efficient mould detection in unhulled paddy.
• The developed methods provide feasible automated recognition solutions for mouldy unhulled paddy.
• This study offers a comparative analysis of various machine learning models for agricultural applications.