Intelligent Transportation Systems (ITS) utilize sensors, cameras, and big data analysis to monitor real-time traffic conditions, aiming to improve traffic efficiency and safety. Accurate vehicle recognition is crucial in this process, and Vehicle Logo Recognition (VLR) stands as a key method. VLR enables effective management and monitoring by distinguishing vehicles on the road. Convolutional Neural Networks (CNNs) have made impressive strides in VLR research. However, achieving higher performance demands significant time and computational resources for training. Recently, the rise of Transformer models has brought new opportunities to VLR. Swin Transformer, with its efficient computation and global feature modeling capabilities, outperforms CNNs under challenging conditions. In this paper, we implement real-time VLR using Swin Transformer and fine-tune it for optimal performance. Extensive experiments conducted on three public vehicle logo datasets (HFUT-VL1, XMU, CTGU-VLD) demonstrate impressive top accuracy results of 99.28%, 100%, and 99.17%, respectively. Additionally, the use of a transfer learning strategy enables our method to be on par with state-of-the-art VLR methods. These findings affirm the superiority of our approach over existing methods. Future research can explore and optimize the application of the Swin Transformer in other vehicle vision recognition tasks to drive advancements in ITS.