Good data drives success, so with Polli, we focus on the sensor and its functionality to achieve the best results and help the drone fly to make that happen. POLLi® can accept and process many different data types (digital color images (RGB), multispectral, thermal, and satellite). Currently, available AI models in POLLi® use color images to count common milkweed or characterize pollinator habitats.

POLLi® algorithms are designed to capture and process imagery quickly and easily with minimal processing. Where possible, RGB imagery is used to simplify the data collection, analysis, and management processes, most of which are automated.

If you have an RGB camera on your drone and want to evaluate flight specifications for your camera, you can find a GSD calculator here. If you have a unique application or need guidance on approaching a unique problem, POLLi® can help you through your computer vision journey.

To simply fly or control the drone's flight path, POLLi® can fly several models and vintages of the DJI brand of drones. Additionally, the Sony Airpeak is also flight planned with POLLi®. Other drones are being added. If a brand of drone you operate isn't available on a current build of the POLLi® Flight application, the specifications of your camera payload can be evaluated, and flight parameters can be provided to you to operate your drone with a different flight controller.

We view data quality and resolution compatibility by the camera rather than the drone model. Depending upon its payload (sensor), each drone is operated to fulfill the data requirements necessary for the best algorithm performance.

Each camera/sensor must fly to capture a 0.25cm Ground Sample Distance (GSD). The camera sensor and lens size can be used to calculate the appropriate elevation above ground level. Follow this link to calculate GSD.

Images must be free of motion blur with good focus and 0.25cm GSD, among other parameters, and must contain an accurate timestamp and coordinates. Appropriate photos may be uploaded into the POLLi® platform and processed through various algorithms. Flight altitude depends on the drone model and camera capabilities (i.e., the DJI Phantom 4 Pro achieves this at 30ft above ground level, while the Sony Airpeak S1 captures 0.25cm and achieves this at 120ft).

POLLi® processes all images for all algorithms individually. The result is the optimum amount of coverage, as efficiently and inexpensively as possible. No image stitching is required.

However, the POLLi® Flight application can be configured to allow your drone to fly and capture at an overlap necessary to process mosaics. Contact POLLi® to get further details on how we can support your current use case.

GCPs are typically used to register stitched imagery to the ground. No, our objective is to quantify habitat, and not necessarily locate the same plant stem repeatedly.

High accuracy location data can be collected using RTK-GPS to achieve sufficient registration accuracy. Note: RTK-GPS uses single images to provide ample geo-location information for most applications. If there's a need to be closer than using RTK-GPS, we can provide access to other drones and payloads to support your registration needs.

This depends on the camera and/or the lens being used. Examples of coverage per image are 0.03 acres/image (DJI Phantom 4 Pro) or 0.10 acres/image (the Sony Airpeak S1 carries Sony's a7R or the ILX-LR1 with a 55mm prime lens).