The purpose of this tool is to generate a set of sample images that closely resemble the types of images that might be taken from a camera mounted on a drone. The images are simulated for approximately 150-300ft of altitude.
In addition, each image is designed to have a wide range of visual characteristics to test a computer vision pipeline's accuracy in real-world conditions. Some of the characteristics include:
- Varying background terrain fidelity, brightness, and contrast
- Varying image angles
- Varying object sizes, shapes, and colors
- Ambient light variations
- Targets of non-interest such as tents, trees, runways markers, roads, people
This tool was made to compete in the SUAS competition
To run the tool, you must install Rust on your machine
- Download Rust from here
- Navigate to the root of the project after cloning it your machine
git clone https://github.com/dec4234/TargetGeneratorV2
- Run
cargo run --releaseto generate the images- Note that the tool will eventually include custom CLI options to give it more flexibility.