RGB image classification with quantum convolutional ansatz

With the rapid growth of qubit numbers and coherence times in quantum hardware technology, implementing shallow neural networks on so-called Noisy Intermediate-Scale Quantum (NISQ) devices has attracted a lot interest. Many (convolutional) circuit ansaetze are proposed for grayscale images classification tasks with promising empirical results. However, when applying these RGB images, intra-channel information that is useful vision not extracted effectively. In this paper, we propose two types to simulate convolution operations which differ way how inter-channel extracted. To best our knowledge, first work convolutional deal effectively, higher test accuracy compared purely classical CNNs. We also investigate relationship between size ansatz learnability hybrid quantum-classical network. Through experiments based CIFAR-10 MNIST datasets, demonstrate larger improves predictive performance multiclass tasks, providing insights near term algorithm developments.