Core Functions

The core functions in mira are for handling data. Object detection data requires working with images and annotations together in a sensible way.

First, we have to track what types of objects we’re detecting, the Categories object stores the list of categories that are permitted as Category objects.

Second, we need to assign detections to images. To do this, we use the Scene object, which consists of an Image, a list of Annotation and an Categories. Each Annotation contains the bounding box and an Category to categorize it.

Finally, we need to group scenes together using a SceneCollection, which also has an Categories assigned to it (the collection has functions for verifying the integrity of its collected scenes). Recognizing this can be confusing, a graph of these relationships is provided below.

graph TD SC[SceneCollection] --> |has many| S[Scene] SC --> |has an| ACF[Categories] S --> |has an| ACF S --> |has an| IM[Image] S --> |has many| A[Annotation] ACF --> |has many| AC[Category] A --> |has an| AC