A Model Integrating Wavelet Convolution and Linear Attention Mechanism for Inventory Management in Power Warehouses

Accurate inventory management in power warehouses is crucial for maintaining consistency between records and actual stock levels. However, the complex storage layout and the diversity of stored materials often lead to human errors during manual inventory checks, resulting in inefficiencies. To address these challenges, we collected 2,000 images from power warehouse environments and created a binary classification dataset containing images of correctly and incorrectly placed items. We propose a novel network, CWAMNet, specifically designed for visual inventory management in complex power warehouse environments. CWAMNet integrates wavelet convolution and a linear attention mechanism to overcome the limitations of traditional convolutional neural networks (CNNs) and standard attention mechanisms, achieving a balance between model capacity and computational efficiency.