Efficient Flower Classification Using EfficientNetB5: A Deep Learning Approach

Recent advances in neural network topologies and the availability of big datasets have led to a significant increase in the use of deep learning models to image classification applications. This study employs the EfficientNetB5 model, a cutting-edge convolutional neural network renowned for its great performance and efficiency, to classify floral species. Five different flower species were used in the training and assessment of the model: tulips, lilies, lotuses, orchids, and sunflowers. Comprehensive metrics and visualizations, such as accuracy and loss curves and confusion matrices, demonstrate the model’s substantial accuracy in categorizing these flower crops, which was made possible by its utilization of the EfficientNetB5 architecture. The outcomes show the EfficientNetB5 model’s potential for practical uses by showing that it can successfully identify between several flower species. The outcomes demonstrate the EfficientNetB5 model’s potential for practical uses in botany research, agricultural monitoring, and automated gardening systems by showing that it can successfully identify several flower species. This paper offers a thorough examination of the model’s performance, identifies the difficulties encountered in categorization, and makes recommendations for future developments. The model performs best at epoch 20, displaying a maximum accuracy of 0.9200 during validation and 0.9227 during training. Similarly, in the training and validation stages, the loss is, respectively, 0.6003 and 0.6006.