FDNet: A Novel Image Focus Discriminative Network for Enhancing Camera Autofocus

• Published in: Neural processing letters Vol. 57; no. 5; p. 76
• Main Authors: Kou, Chenhao, Xiao, Zhaolin, Jin, Haiyan, Guo, Qifeng, Su, Haonan
• Format: Journal Article
• Language: English
• Published: New York Springer US 18.08.2025; Springer Nature B.V
• Subjects: Ablation; Accuracy; Annotations; Artificial Intelligence; Cameras; Classification; Complex Systems; Computational Intelligence; Computer Science; Datasets; Decision making; Efficiency; Feature extraction; Genetic algorithms; Image quality; Lighting; Optimization; Robustness; Semantics; Focus discrimination; Genetic algorithm; Coordinate attention; Multi-scale feature fusion; Autofocus
• ISSN: 1573-773X; 1370-4621; 1573-773X
• DOI: 10.1007/s11063-025-11788-0

Accurate activation and optimization of autofocus (AF) functions are essential for capturing high-quality images and minimizing camera response time. Traditional contrast detection autofocus (CDAF) methods suffer from a trade-off between accuracy and robustness, while learning-based methods often incur high spatio-temporal computational costs. To address these issues, we propose a lightweight focus discriminative network (FDNet) tailored for AF tasks. Built upon the ShuffleNet V2 backbone, FDNet leverages a genetic algorithm optimization (GAO) strategy to automatically search for efficient network structures, and incorporates coordinate attention (CA) and multi-scale feature fusion (MFF) modules to enhance spatial, directional, and contextual feature extraction. A dedicated focus stack dataset is constructed with high-quality annotations to support training and evaluation. Experimental results show that FDNet outperforms mainstream methods by up to 4% in classification accuracy while requiring only 0.2 GFLOPs, 0.5 M parameters, a model size of 2.1 MB, and an inference time of 0.06 s, achieving a superior balance between performance and efficiency. Ablation studies further confirm the effectiveness of the GAO, CA, and MFF components in improving the accuracy and robustness of focus feature classification.