Enhanced Object Detection Using Efficient YOLOv8 Fractional Feedforward Network and Spider-Tailed Horned Viper Optimization

Object detection systems often struggle with challenges such as noise, illumination variation, and partial occlusion, which degrade both localization accuracy and recognition performance. To address these issues, this study introduces a novel Fractional You Only Look Once Efficient Network (Frac-YOLOEffNet), which integrates Fractional Concept (FC) with the YOLOv8 and EfficientNet architectures to improve feature sensitivity and robustness. The initial step involves applying an Adaptive Median Filter to the input image for noise reduction. Moreover, extraction of features is performed, where the significant features, such as Shape Local Binary Texture (SLBT), Gray-Level Co-occurrence Matrix (GLCM), Speeded Up Robust Features (SURF), Oriented Fast and Rotated BRIEF (ORB), and Scale-Invariant Feature Transform (SIFT), are extracted. Object recognition is later achieved with a Deep Convolutional Neural Network (DCNN) with weights and bias trained via the Spider-Tailed Horned Viper Optimization (STHVO). The developed Frac-YOLOEffNet model achieved an F1-score of 96.168%, a recall of 97.877%, and an accuracy of 96.998%, respectively.