How can I preserve tables, images, and nested lists when converting DOCX WORD to Markdown?

Part of

Saliency-Based Visual Attention Models