Rotten and Fresh Fruits Classification using Deep Learning

Abstract

Accurate and timely classification of fresh and decaying fruits is crucial for upholding quality standards in the food industry. This study introduces a deep learning approach that utilizes images to distinguish between fresh and rotten fruits. The proposed solution effectively addresses this classification challenge by control the capabilities of Convolutional Neural Networks. In this research a diverse dataset is used, comprising labeled images of various fruits in varying states of freshness, encompassing both rotten and fresh examples. Key preprocessing steps, including resizing and normalizing the dataset, is implemented to ensure that the model receives a uniform input. The core of the methodology involves initializing a pre-trained CNN architecture. This approach enables the adaptation of the pretrained model to the specific fruit classification task, ultimately providing an efficient and effective solution for quality control in the food sector. Through the utilization of this approach, our model can inherit valuable feature representations from its pretrained counterpart. To evaluate the performance of the trained model, a distinct test dataset is employed and the results encompass key measurements of performance. With an impressive accuracy of 97.5%, the proposed model outperforms the base paper, which achieves 96.05%. This advancement highlights the model’s proficiency in distinguishing between fresh and rotten fruits, underscoring its practical effectiveness.

Topics & Concepts