We are an image recognition company, enabling automated object counting on aerial imagery.
We can recognize and count many object types, like trees, cars, or certain animals. To use the app, you just have to upload the pictures made with from drone (or an airplane) - they are automatically converted to an orthomosaic map and analysed by the app. To improve our model, we add new object types and customise it for our clients. In order to do that, we get pictures of new objects from users, label them manually, then train the model on this data set and it’s ready for automated counting.
As far as we know, there is only one other company that works on automated image recognition from drones/planes and its technology is much worse than ours. The indirect competitor are companies that develop tools to analyse satellite imagery, however our market is too niche for them. We already have 300+ users since launching the app three months ago and we had no marketing - they are finding us every day. (different needs - agriculture, insurance, forestry, etc.)
The problem we can solve automatically is currently either done manually from ground level (e.g. forest inspections) or done manually from drone/plane imagery - there are businesses in the third world which pay 50 cents - 1 dollar per hour for human labelling of pictures. We have developed a neural network that replicates the way human vision recognizes certain objects and does it quickly and automatically. Our team consists of a winner of Google image recognition challenge.