utils/modes.py

OPEN_VOCABULARY_DETECTION = "open vocabulary detection + masks"
CAPTION_GROUNDING_MASKS = "caption + grounding + masks"

INFERENCE_MODES = [
    OPEN_VOCABULARY_DETECTION,
    CAPTION_GROUNDING_MASKS
]