Tight Pants Occlusion Image Dataset

The retail e-commerce industry faces significant challenges in accurately detecting clothing items due to occlusions caused by overlapping garments, leading to potential misclassifications and poor customer experience. Existing datasets often lack sufficient examples of occluded items, limiting the effectiveness of machine learning models. This dataset aims to address these shortcomings by providing a comprehensive collection of images that showcase tight pants being obstructed by various clothing items, facilitating better training and evaluation of object detection systems. The data is collected using high-resolution cameras in diverse retail environments, ensuring a wide variety of poses and lighting conditions. Quality control measures include multi-round annotations and expert reviews to ensure the highest level of annotation accuracy. The dataset is organized in a standard JPG format, with images stored in structured directories for easy access.