The availability of large and diverse training datasets is critical for the successful implementation of machine learning (ML) and deep learning (DL) models across various domains. In agriculture, applications such as plant classification and disease detection depend heavily on specific crop types, requiring extensive datasets to enhance accuracy (Khaire et al. in Proceedings of the 2023 14th international conference on computing communication and networking technologies (ICCCNT). IEEE, 2023; Isola et al. in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017). Obtaining such datasets is challenging due to variations in climatic conditions and lighting environments. To address this limitation, image-to-image (I2I) translation has gained prominence as a data augmentation technique, widely used in image segmentation, annotation, enhancement, and style transfer (Gnd in Tensor-to-image: image-to-image translation with vision transformers, 2021). The technique of image-to-image (I2I) translation has gained popularity in recent years due to its versatility in applications such as image segmentation, data annotation, image enhancement, synthesis, pose prediction, and style transfer in computer vision and image processing. Various deep learning algorithms, including RNNs, CNNs, GANs, among others, have been employed for I2I translation. Additionally, transformers have emerged as a viable approach for this purpose. This work delves into the application of vision transformer (ViT) for unpaired image translation, focusing on three distinct datasets: ImageNet, Cityscape, and Soybean Crop. The effectiveness of deep learning models largely depends on the availability of diverse and abundant training data. However, in domains like agriculture, acquiring such datasets is challenging due to variations in plant types, environmental conditions, and lighting. This study explores the use of Vision Transformers (ViT) for unpaired image-to-image (I2I) translation, a technique widely applied in tasks such as image segmentation, style transfer, and data augmentation (Khaire et al. in Proceedings of the 2023 14th international conference on computing communication and networking technologies (ICCCNT). IEEE, 2023, Gnd in Tensor-to-image: image-to-image translation with vision transformers, 2021). Unlike traditional deep learning models such as convolutional neural networks (CNNs) and generative adversarial networks (GANs), ViTs offer superior feature extraction capabilities (Isola et al. in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017). This research evaluates the performance of ViT on three datasets: ImageNet (Kim et al. in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022), Cityscape (Mignoni in Data Brief 40:107756, 2022), and Soybean Crop (Nazki et al. in Comput Electr Agricult 168:105117, 2020). The goal is to assess ViT’s adaptability to different visual domains and compare its effectiveness against existing methods.