Scarf Occlusion Image Dataset

The retail e-commerce industry is experiencing rapid growth, yet it faces significant challenges in accurate product recognition due to occlusions in images, such as clothing items covering parts of the product. Existing datasets often lack sufficient representation of these scenarios, leading to poor model performance in real-world applications. This dataset aims to address the need for high-quality images that depict scarves occluding parts of products, enhancing the training of models for better recognition accuracy. The data is collected through various online retail platforms, capturing diverse occlusion scenarios in natural settings. Quality control measures include multi-round annotations, consistency checks, and expert reviews to ensure high accuracy and reliability. The data is stored in JPG format, organized systematically by categories of occlusion types for easy access.