Efficient Classification of Rotten and Healthy Bell Peppers Using an Optimized DenseNet121 Model

In the countries like India with the growing population and rising food demands, agriculture, cold storage and transportation process plays a critical role in the country, highlighting the need to maximize crop yields and food preservation. Bell pepper cultivation and classification of rotten and healthy bell peppers during preservation in cold storage and transportation process is severely affected by the deterioration of produce, which results in substantial financial losses. To minimize wastage and ensure the quality of produce, timely and accurate classification of rotten versus healthy bell peppers is important. With the rise of computer vision and machine learning, automated classification systems offer a promising solution. The study proposes a novel deep-learning method for effectively classifying rotten bell peppers and healthy bell peppers using the optimized DenseNet121 model, which enhances the accuracy of classification. Due to the optimized DenseNet121 model's exceptional performance in feature extraction and classification, it offers a practical solution that can be utilized not only in cultivation fields but also in storage areas where bell peppers are kept before they are sent to factories. This can substantially improve the process of sorting, reducing post-harvest losses, and supporting productivity in agriculture by ensuring only high-quality supply moves further along the supply chain. Experimental results show that the DenseNet121 model achieves a remarkable 98.81% accuracy in distinguishing healthy and rotten bell peppers.