Google Maps Platform Launches Three AI Image Features: Street View Ground Truth Anchoring, Satellite Imagery Auto-Annotation, Pre-trained Land Cover Classification Model

According to OneMillion AI's monitoring, Google Maps Platform announced three AI imagery updates at Cloud Next. The Maps Imagery Grounding feature allows users to input a text prompt and generate an AI image with Google Street View imagery as the background, suitable for advertising creative and film previews. For example, entering "a futuristic spaceship hovering in front of the Washington Square Arch," the system, within seconds in the Gemini Enterprise Agent Platform, outputs a synthetic image based on a real location, which can also be transformed into a video using Veo. This feature is currently available in the United States in a Private Preview mode, with the advertising group WPP currently testing it.

The Aerial and Satellite Insights is a new aerial and satellite imagery dataset in Google Earth AI, which, when accessed through BigQuery, can automatically analyze image content, reducing the need for weeks of manual review to a matter of minutes. Use cases include monitoring city construction sites and assisting in infrastructure planning.

Simultaneously, two Earth AI Imagery models have been launched, available on Google Cloud Model Garden in an Experimental state. These models are pre-trained to recognize features such as bridges, roads, and power lines, allowing users to use them in their products without the need for training from scratch. The geospatial intelligence company Vantor has integrated these models into its Sentry application to convert post-disaster satellite images into reports identifying damaged infrastructure.