Hybrid Generative Approaches for Flower Classification

Various techniques have been employed in the development of image classification frameworks to enhance efficiency and accuracy. In this work, flower image classification is performed using multiple approaches, including edge detection and feature extraction combined with the K-Nearest Neighbor (KNN) algorithm. A Deep Convolutional Neural Network (CNN) is also integrated with traditional machine learning classifiers to improve the identification of flower species. To further enhance accuracy, we propose four distinct hybrid frameworks for flower classification. The dataset, sourced from Kaggle, includes five flower categories: rose, tulip, dandelion, sunflower, and daisy. Each model is evaluated using accuracy, precision, and F-score metrics. The experimental results show that the hybrid model combining ResNet with Support Vector Machine (SVM) achieves the highest accuracy among all proposed models. Overall, the study demonstrates the effectiveness of hybrid fusion-based classification techniques for flower image recognition.