Chocolate Product Occlusion Image Dataset

The retail e-commerce industry faces challenges in accurately identifying products that are partially occluded in images. Existing datasets often lack diversity in occlusion scenarios, leading to poor performance in real-world applications. This dataset aims to address the need for a comprehensive collection of images depicting chocolate products with various levels of occlusion. The dataset was collected using high-resolution cameras in controlled lighting environments, ensuring clarity and detail. Quality control measures included multi-round annotations, consistency checks among annotators, and expert reviews to ensure accuracy. Images are stored in JPG format, organized by product type and occlusion level for easy access and analysis. This dataset's core advantages lie in its high-quality annotations, which boast over 95% accuracy and consistency, addressing the common issues of mislabeling and incomplete data. The innovative annotation process incorporates advanced techniques such as semi-automated labeling and machine learning-assisted verification, significantly reducing human error. The dataset is designed to enhance performance in real-world applications, with expected improvements in product recognition accuracy by over 20% in scenarios involving occluded products compared to existing datasets.