A lightweight convolutional neural network-based feature extractor for visible images

• Published in: Computer vision and image understanding 2024-12, Vol.249, p.104157, Article 104157
• Main Authors: He, Xujie, Jin, Jing, Jiang, Yu, Li, Dandan
• Format: Article
• Language: English
• Subjects: Computer vision; Feature extraction; Feature extraction network; Lightweight; Memory efficiency
• ISSN: 1077-3142
• DOI: 10.1016/j.cviu.2024.104157

Feature extraction networks (FENs), as the first stage in many computer vision tasks, play critical roles. Previous studies regarding FENs employed deeper and wider networks to attain higher accuracy, but their approaches were memory-inefficient and computationally intensive. Here, we present an accurate and lightweight feature extractor (RoShuNet) for visible images based on ShuffleNetV2. The provided improvements are threefold. To make ShuffleNetV2 compact without degrading its feature extraction ability, we propose an aggregated dual group convolutional module; to better aid the channel interflow process, we propose a γ-weighted shuffling module; to further reduce the complexity and size of the model, we introduce slimming strategies. Classification experiments demonstrate the state-of-the-art (SOTA) performance of RoShuNet, which yields an increase in accuracy and reduces the complexity and size of the model compared to those of ShuffleNetV2. Generalization experiments verify that the proposed method is also applicable to feature extraction tasks in semantic segmentation and multiple-object tracking scenarios, achieving comparable accuracy to that of other approaches with more memory and greater computational efficiency. Our method provides a novel perspective for designing lightweight models. •An accurate and lightweight feature extractor is proposed, termed RoShuNet.•The aggregated dual group convolutional (A-DGC) module used for feature extraction is proposed.•The γ-weighted shuffle (γ-WSM) module employed for aiding in channel interflow is proposed.•Slimming strategies addressing dual group convolutional and short-cut structures are introduced.•Extensive experiments demonstrate the SOTA performance of the proposed method.