Canvas Shoe Occlusion Image Set

The retail e-commerce industry faces significant challenges regarding accurate product recognition due to occlusions that can mislead visual search algorithms. Current solutions often struggle with occluded images, leading to lower accuracy in product identification and customer dissatisfaction. This dataset aims to address these issues by providing a robust collection of images featuring canvas shoes with various occlusions such as sock coverage, laces obstructing logos, and side views that diminish shoe shape visibility. The data collection involved capturing images in controlled environments using high-resolution cameras, ensuring clear visibility of occlusions. Quality control measures, including multiple rounds of annotation by trained professionals and consistency checks, were implemented to maintain high standards. The dataset is structured in JPG format, organized with clear metadata for each image, facilitating easy access and usability for training machine learning models.