Inc-ghostNet: A Lightweight Muti-Branch Ghost Module Fusion Network for Image Classification

The high computational cost and large model size of traditional convolutional neural networks (CNNs) present significant challenges for their deployment on resource-constrained devices, such as mobile devices, embedded systems and IoT devices. In order to address these challenges, we propose Inc-ghostNet, a lightweight multi-branch network that combines Inception structures, Ghost modules, and DFC attention mechanisms for the purpose of efficient image classification. The experimental results demonstrate that, in comparison with InceptionNeXt and other classic lightweight networks, Inc-ghostNet exhibits notable advantages in terms of classification accuracy, model parameters, volume, and computational complexity.