Zhejiang Deploys 24/7 AI Computer Vision to Regulate 190,000 Takeout Kitchens
Zhejiang Province has deployed an automated 'AI+ Off-site Regulation' system that continuously inspects and samples more than 190,000 food delivery merchants across the province using computer vision cameras. The platform automatically detects hygiene violations, calculates dynamic risk scores, and pushes real-time evidence and warnings to both merchants and regulatory officers. This deployment marks a massive real-world application of computer vision in public health enforcement and smart governance. By integrating camera analytics with electronic licensing databases across major food delivery platforms, the system helps eliminate unlicensed 'ghost kitchens' and improves overall food safety efficiency. The system dynamically captures 12 distinct violation types, including rodent activity at night, stray animals in kitchens, mops washed in vegetable sinks, and staff not wearing sanitary caps. Merchants scoring under 60 points receive automated SMS and AI calls to self-correct, with unrectified cases escalating to physical inspections, which has raised merchant self-correction rates from ~20% to 60%.
## BACKGROUND
'Ghost kitchens' refer to delivery-only food vendors that operate with fake registered addresses, lack valid licenses, or maintain unsanitary working conditions hidden from public view. In response, Chinese market regulators are promoting 'off-site regulation' to shift food safety enforcement from labor-intensive manual inspections to automated, continuous digital monitoring.