Create a vendor selection project & run comparison reports
Click to express your interest in this report
Indication of coverage against your requirements
A subscription is required to activate this feature. Contact us for more info.
Celent have reviewed this profile and believe it to be accurate.


CrowdAI combines deep learning with available imagery to detect, classify, and monitor assets.

Most things in the world are influenced by nearby assets. Historically, institutions have used manual processes paired with aerial, drone, and satellite imagery to understand the world around them. The problem is that the process to measure the relationship of those assets is costly, time-intensive, and susceptible to inaccuracy.

CrowdAI empowers organizations to make operational decisions by forecasting changes in the physical world.

Clients encompassing ride-sharing, government agencies, telecommunications, utilities, and insurance trust CrowdAI to identify growth opportunities, optimize business operations, and mitigate risk.

CrowdAI is built by former Google Maps and IBM Watson engineers and backed by Y Combinator.

Why CrowdAI?

- Data Security - On premise/Cloud deployed. PII data stays internal.

- Geographic scale - Classify, detect, and monitor objects over large Areas of Interest

- Speed - Classify, detect, and monitor thousands of objects of an entire city in a matter of minutes

- International - CrowdAI's models are trained to be globally flexible based on available imagery regardless of region

- Customization - Proprietary models per use case - per Customer; trained on Customer’s data and specific object needs.

- Third party validation - Won highly competitive bids with US government and publish peer-reviewed white papers

- Imagery agnostic - Quickly analyze data regardless of the source of the imagery

- Platform agnostic - Integrate into existing workflows and tools. Deliver data in the most convenient format for the Customer (API, FTM, CSV, s3, geojson, etc).

- Maturity - Deep-learning models have been training for 3.5 years across 125 countries, multiple biomes and different imagery types. CrowdAI creates all the training data. Trains models in 2.5 weeks or less.

- Value - 100% deep learning technology limits the need for human labor to produce deliverables. Reduces costs to the Customer.

Geographic Range
CrowdAI's models are trained to be globally flexible and can be fine-tuned on imagery from commercial and proprietary satellite, manned aerial, and unmanned aerial sources. The geographic range is relatively unlimited, as satellite constellations image the landmass of the earth. There is one important caveat: most areas of the open ocean are not imaged regularly, and therefore are not typically available.

Resolution Quality
CrowdAI has strong partnerships with all the major third-party imagery providers including (but not limited to) Digital Globe, Airbus, Nearmap, NOAA, and Planet. Different objects have different resolution quality based on available imagery sources. All of these are commercially available over major populated areas. Generally speaking, the entire Continental United States is covered usually two to three times per year between all providers.

Resolution varies from 5.8cm/pixel which is detailed enough to see a crack on a sidewalk to 3m/pixel which is detailed enough to know if a building does or does not exist. In addition, post-CAT, imagery is often sourced from 25cm/pixel aerial data.