EFTViT: Efficient Federated Training of Vision Transformers with Masked Images on Resource-Constrained Edge Devices