This repository contains the Unity project used to generate data for CV model training.
Unity Version: 2023.1.16 (any version of 2023.1.x should work) Perhaps someone can try loading this in a later version and see if it works.
- Clone this repository
- Open Unity Hub, and install a version of Unity if not done already.
- On the top, click add, then select the root directory of this repository on your local device.
- Once the project is opened, ensure you are in the Data_Gen scene (shown in the hierarchy). If not, navigate to Assets/Scenes in the project tab, and open Data_Gen
- Click the button on top to run the program to generate data
- Access the images/annotations by accessing RotationPoint/Main Camera in the hierarchy, then go to the Percpetion Camera object, scroll down and click "Show Folder" on the Latest Generated Dataset
Accessing the annotations is a bit of a pain (currently we use an external python script to organize everything.
Scripts are found in Assets/Scripts.
Simulation Scenario: This object is responsible for starting the process of spawning the objects. This object relies on WaterObjectRandomizer.cs, and its parameters can be modified within the object. WaterObjectRandomizer.cs: The main script associated with placing the objects and orienting the camera. Main Camera (in RotationPoint): Visualizes boudning boxes and saves image data.
- Initial UI before starting generation doesn't work