Sneaker Occlusion Image Dataset

In the current retail e-commerce landscape, visual recognition systems face significant challenges due to occlusions affecting product visibility, which can lead to misclassification and poor user experience. Existing datasets often lack diverse occlusion scenarios, limiting the training effectiveness of machine learning models. This dataset aims to address these challenges by providing a comprehensive collection of images where sneakers are obscured by factors such as pant legs, blurred shoe tips, and ground reflections or mud. The data is collected using high-resolution cameras in various retail environments to ensure realistic conditions. Quality control measures include multiple rounds of annotation, consistency checks among annotators, and expert reviews to ensure high accuracy. The dataset is organized in JPG format, each image accompanied by metadata in a structured manner.