| Mobile Web

Nvidia unveils fast object-detection AI that reads photos, UI and documents

Nvidia has unveiled LocateAnything, an AI model designed to rapidly detect objects in photos and screenshots and identify UI elements and text locations in applications and documents. The company describes it as a vision-language model specialised for high-speed detection. Nvidia says demonstrations show fast identification and finer distinctions than Qwen3-VL and REX-Omni, including repeated objects and text recognition. It released the model openly via Hugging Face and provides a separate demo app.