Improving Mango Quality Assessment: A Multi-Layer Perceptron Approach for Grading

• Published in: 2024 2nd World Conference on Communication & Computing (WCONF) pp. 01 - 04
• Main Authors: Kaur, Arshleen, Sharma, Rishabh, Thapliyal, Nitin, Aeri, Manisha
• Format: Conference Proceeding
• Language: English
• Published: IEEE 12.07.2024
• Subjects: Accuracy; Adaptation models; Agricultural Produce Grading; Agricultural products; Artificial Neural Networks; Automated Quality Control; Data models; Fruit Quality Classification; Machine learning; Machine Learning in Agriculture; Mango Quality Assessment; Multi-Layer Perceptron; Performance evaluation; Quality assessment; Quality control; Training
• DOI: 10.1109/WCONF61366.2024.10692009

The quality standard of an agricultural product, which is often mango, one of the most popular foods in my country, performs a critical function of selecting the best from the market and providing the highest satisfaction to the consumers. Checking the quality for a long time is still difficult because it uses manual inspection and simple mechanical devices, which can't ensure that the evaluation is consistent, objective, and efficient. This study comes up with a new method of mango quality evaluation that relies on a Multi-Layer Perception (MLP) network-the most powerful artificial neural network with the ability to perform pattern recognition and classification tasks well. The model was developed and trained on a dataset comprising both external and internal quality attributes of mangoes, classified into four quality levels: The Genius category is the highest one, followed by the Excellent, the Good, and the last category Fair. MLP showed excellent performance in this aspect, achieving 95.7% general classifiable precision, which far exceeds the traditional methods with the highlighting of machine learning as the method that could push the assessment of fruit quality to the other level. This study, not only, involved the investigation of the feasibility and efficiency of MLP models for the evaluation purpose of agricultural quality, but also, it proposed new directions for future studies such as data of sensory influences and broadening of scope to other products of agriculture. The report points to one of the most important positive aspects of the experiment in terms of a more trustworthy assessment of quality with further selection of the producers, dealers, and buyers in the agricultural chain.