Midv-74 -

1000 upright and 1000 rotated scans of the documents. Photos: 1000 high-resolution photos of the documents.

The dataset is utilized for various tasks within the field of document analysis and recognition: 1. Document Detection and Localization midv-74

The rapid digitalization of financial and government services has made the automatic processing of identity documents (IDs) an essential technology. From opening bank accounts on mobile apps to boarding flights, Automated Teller Machines (ATMs) and remote Know Your Customer (KYC) processes depend on robust Optical Character Recognition (OCR) and document analysis. However, training these AI systems requires massive, diverse, and annotated datasets—a rarity due to strict privacy regulations and security restrictions. 1000 upright and 1000 rotated scans of the documents

The datasets are created by Smart Engines to fill the scarcity gap in public document datasets. The datasets are created by Smart Engines to

MIDV-74 has garnered attention for its portrayal of a lesser-known aspect of Soviet history and its exploration of the human side of intelligence operations.

Designed to overcome the lack of variability in previous datasets by providing unique, synthetically generated data, comprising 1000 unique mock documents and over 72,000 annotated images. MIDV-2020 Dataset Overview