Abstract: This work introduces a CLIP-guided Masked Autoencoder (CGD-MAE) pre-training strategy designed to enhance the performance of existing ViT-based architectures for vehicle re-identification (V ...