M5Product Image Classification Dataset

M5Product is a large-scale multimodal pre-training dataset with coarse-grained and fine-grained annotations for electronic products.

• Consists of over 6 million multimodal pairs with over 5,000 attributes.
• Contains rich multimodal information, including images, text, tables, videos, and audio.
• Contains 6 million category annotations, covering more than 6,000 categories.
• Extensive data sources (provided by 1 million merchants).