Introduction

The Capsule class provides information to the application about what a capsule is and how it should be run. Every capsule defines a Capsule class that extends BaseCapsule in its capsule.py file.

from vcap import (
    BaseCapsule,
    NodeDescription,
    options
)

class Capsule(BaseCapsule):
    name = "detector_person"
    version = 1
    stream_state = StreamState
    input_type = NodeDescription(size=NodeDescription.Size.NONE),
    output_type = NodeDescription(
        size=NodeDescription.Size.ALL,
        detections=["person"])
    backend_loader = backend_loader_func
    options = {
        "threshold": options.FloatOption(
            default=0.5,
            min_val=0.1,
            max_val=1.0)
    }

Fields

name

name: str

The name of the capsule. This value uniquely defines the capsule and cannot be shared by other capsules. By convention, the capsule's name is prefixed by some short description of the role it plays ("detector", "recognizer", etc) followed by the kind of data it relates to ("person", "face", etc) and, if necessary, some differentiating factor ("fast", "close_up", "retail", etc).

version

version: int

The incremental version of the capsule. It's recommended to change the version on any change of capsule options, algorithms or models.

device_mapper

device_mapper: DeviceMapper

A device mapper contains a single field, filter_func, which is a function that takes in a list of all available device strings and returns a list of device strings that are compatible with this capsule.

stream_state

stream_state: Type[BaseStreamState]

(Optional) A class that the application will make instances of for each video stream. See the section on StreamState for more information.

input_type

input_type: NodeDescription

Describes the types of DetectionNodes that this capsule takes in as input. See the section on inputs and outputs for more information.

output_type

output_type: NodeDescription

Describes the types of DetectionNodes that this capsule produces as output. See the section on inputs and outputs for more information.

backend_loader

backend_loader: Callable[[dict, str], BaseBackend]

A function that creates a backend for this capsule. Takes the following as arguments:

• capsule_files: Provides access to all files in the capsule. The keys are file names and the values are bytes.
• device: A string specifying the device that this backend should use. For example, the first GPU device is specified as "GPU:0".

The function must return a class that subclasses BaseBackend. See the section on backends for more information.

options

options: dict[str, Option]

A dict where the key is an option name and the value is an Option object. See the section on options for more information.