T-Shirt Try-On Composite Image Dataset

The current retail e-commerce industry is facing increasingly fierce competition, and consumers have a growing demand for personalized shopping experiences. However, traditional try-on methods cannot meet users' expectations, especially in online shopping, where consumers often cannot intuitively experience the effect of clothing. Existing virtual try-on technologies mostly lack realism and accuracy, making it difficult to provide users with satisfactory shopping experiences. This dataset aims to create high-quality synthetic T-shirt try-on images to address user confusion during selection and increase purchase conversion rates. The dataset is constructed using advanced image processing techniques, combined with real users' try-on data, and generates composite images through multiple iterative rounds. Data collection is conducted using professional photographic equipment under different lighting and background environments to ensure image quality. Quality control involves multiple rounds of annotation, consistency checks, and expert reviews to ensure data accuracy and reliability. Ultimately, the data is stored in JPEG format and organized into a correspondence between user IDs and composite images to facilitate subsequent retrieval and analysis.