DevilCV: Computer Vision service for VEXBridge

DevilCV is a computer vision service for VEXBridge. It provides an interface to detect colors using OpenCV and send them to the Bridge.

Detection

The detection data sent to VEXBridge is made up of a center (center_x) and an edge (edge_x).

The full detection type definition is as follows:

class Detection(BaseModel):
    bounding_box: cv2.typing.Rect # tuple of (x, y, w, h)
    center: Center

Meaning of detection values

Detection values represent relative positions of the detected object from the center of the camera, where the center is 1, the leftmost point is 0.5 (meaning 50% of the way to the center), and the rightmost point is 1.5 (meaning 50% over the center).

Value Paths

Paths always start with vision, followed by the detection type, then class type, and finally the variable name. For example, a red color detector will have paths vision/color/red/center_x and vision/color/red/edge_x.

Installation

Clone the repository and install the dependencies using pip:

git clone https://github.com/UWST-Robotics/DevilCV
cd DevilCV
pip install -r requirements.txt

.env File

Create a .env file in the root directory of the project. This file should contain the following variables:

VEXBRIDGE_HOST: str # IP address or hostname of the VEXBridge server
VEXBRIDGE_PORT: int # port of the VEXBridge server
DEVICE: "pi" | "windows" # device type for capture mode
CAMERA_EXPOSURE: int # camera exposure 

Running the Service

To run the service, use the following command:

python -m DevilCV.main