Mobiusi/Backpack-Occlusion-Image-Dataset

Backpack Occlusion Image Dataset

In the current retail e-commerce sector, visual recognition systems face significant challenges in accurately identifying products due to occlusions, particularly in images where straps obscure logos or buckles. Existing datasets often lack diversity in occluded images, making them insufficient for training robust models. This dataset aims to address these limitations by providing a specialized collection of images featuring backpacks with varying levels of occlusion caused by shoulder straps. The dataset is collected using high-quality cameras in diverse environments, ensuring a variety of angles and lighting conditions. Quality control measures include multi-round annotations and expert reviews to ensure consistency and accuracy. The images are stored in JPG format, organized by occlusion level and category for easy access.

Technical Specifications

Field	Type	Description
file_name	string	File name
quality	string	Resolution
backpack_presence	boolean	Whether a backpack is present in the image.
backpack_position	string	The approximate position where the backpack appears in the image, such as 'top left', 'bottom right', etc.
occlusion_level	string	The extent to which the backpack is obstructed in the image, such as 'partially obstructed' or 'fully obstructed'.
background_clutter	string	The level of complexity of the background, such as 'simple', 'medium', or 'complex'.

Compliance Statement

Authorization Type	CC-BY-NC-SA 4.0 (Attribution–NonCommercial–ShareAlike)
Commercial Use	Requires exclusive subscription or authorization contract (monthly or per-invocation charging)
Privacy and Anonymization	No PII, no real company names, simulated scenarios follow industry standards
Compliance System	Compliant with China's Data Security Law / EU GDPR / supports enterprise data access logs

Source & Contact

If you need more dataset details, please visit Mobiusi. or contact us via contact@mobiusi.com