Vision Reference

kevinbotlib.vision

kevinbotlib.vision.SingleFrameSendable

Bases: BinarySendable

Sendable for a single frame of video or an image

encoding instance-attribute

Frame encoding format

Supported encodings: * JPG * PNG

data_id = 'kevinbotlib.vision.dtype.frame' class-attribute instance-attribute

Internally used to differentiate sendable types

struct = {'dashboard': [{'element': 'value', 'format': 'limit:1024'}, {'element': 'resolution', 'format': 'raw'}, {'element': 'quality', 'format': 'raw'}, {'element': 'encoding', 'format': 'raw'}]} class-attribute

Data structure suggestion for use in dashboard applications

get_dict()

Return the sendable in dictionary form

Returns:

Name	Type	Description
dict	dict	The sendable data

kevinbotlib.vision.MjpegStreamSendable

Bases: SingleFrameSendable

Sendable for a single frame of an MJPG stream

Contains all required information for decoding a video stream

data_id = 'kevinbotlib.vision.dtype.mjpeg' class-attribute instance-attribute

Internally used to differentiate sendable types

quality instance-attribute

The current JPEG compression level out of 100 - lower means more compression

resolution instance-attribute

A two integer list containing the video resolution (WIDTH x HEIGHT)

encoding = 'JPG' class-attribute instance-attribute

Frame encoding format

struct = {'dashboard': [{'element': 'value', 'format': 'limit:1024'}, {'element': 'resolution', 'format': 'raw'}, {'element': 'quality', 'format': 'raw'}, {'element': 'encoding', 'format': 'raw'}]} class-attribute

Data structure suggestion for use in dashboard applications

get_dict()

Return the sendable in dictionary form

Returns:

Name	Type	Description
dict	dict	The sendable data

kevinbotlib.vision.FrameEncoders

Encoders from OpenCV Mats into raw bytes or network sendables

encode_sendable_jpg(frame, quality=80) staticmethod

Encode an OpenCV Mat to a SingleFrameSendable using JPEG encoding

Parameters:

Name	Type	Description	Default
frame	MatLike	The Mat to encode	required
quality	int	The JPEG quality level. Defaults to 80.	80

Returns:

Name	Type	Description
SingleFrameSendable	SingleFrameSendable	A sendable to be sent over the network

encode_sendable_png(frame, compression=3) staticmethod

Encode an OpenCV Mat to a SingleFrameSendable using PNG encoding

Parameters:

Name	Type	Description	Default
frame	MatLike	The Mat to encode	required
compression	int	The PNG compression level. Defaults to 3.	3

Returns:

Name	Type	Description
SingleFrameSendable	SingleFrameSendable	A sendable to be sent over the network

encode_jpg(frame, quality=80) staticmethod

Encode an OpenCV Mat to raw bytes using JPEG encoding

Parameters:

Name	Type	Description	Default
frame	MatLike	The Mat to encode	required
quality	int	The JPEG quality level. Defaults to 80.	80

Returns:

Name	Type	Description
bytes	bytes	Raw data

encode_png(frame, compression=3) staticmethod

Encode an OpenCV Mat to raw bytes using PNG encoding

Parameters:

Name	Type	Description	Default
frame	MatLike	The Mat to encode	required
compression	int	The PNG compression level. Defaults to 3.	3

Returns:

Name	Type	Description
bytes	bytes	Raw data

kevinbotlib.vision.FrameDecoders

Decoders from Base64 or network sendables to OpenCV Mats

decode_sendable(sendable) staticmethod

Decode a SingleFrameSendable into an OpenCV Mat

Parameters:

Name	Type	Description	Default
sendable	SingleFrameSendable	The sendable to decode	required

Raises:

Type	Description
ValueError	If the encoding type isn't recognized

Returns:

Name	Type	Description
MatLike	MatLike	An OpenCV Mat

decode_base64(data, encoding) staticmethod

Decode a base64 string into an OpenCV Mat

Parameters:

Name	Type	Description	Default
data	str	The base64 data to decode	required
encoding	str	The encoding format. Can be JPG or "PNG"	required

Raises:

Type	Description
ValueError	If the encoding type isn't recognized

Returns:

Name	Type	Description
MatLike	MatLike	An OpenCV Mat

kevinbotlib.vision.VisionCommUtils

Various utilities to integrate vision data with networking

init_comms_types(client) staticmethod

Allows the use of frame data over the communication client

Parameters:

Name	Type	Description	Default
client	RedisCommClient	The communication client to integrate with	required

kevinbotlib.vision.BaseCamera

Bases: ABC

Abstract class for creating Vision Cameras

get_frame() abstractmethod

Get the current frame from the camera

Returns:

Type	Description
tuple[bool, MatLike]	tuple[bool, MatLike]: Frame retrieval success and an OpenCV Mat

set_resolution(width, height) abstractmethod

Attempt to set the current camera resolution

Parameters:

Name	Type	Description	Default
width	int	Frame width in px	required
height	int	Frame height in px	required

kevinbotlib.vision.CameraByIndex

Bases: BaseCamera

Create an OpenCV camera from a device index

Not recommended if you have more than one camera on a system

__init__(index)

Initialize the camera

Parameters:

Name	Type	Description	Default
index	int	Index of the camera	required

kevinbotlib.vision.CameraByDevicePath

Bases: BaseCamera

Create an OpenCV camera from a device path

__init__(path)

Initialize the camera

Parameters:

Name	Type	Description	Default
path	str	Device path of the camera ex: /dev/video0	required

kevinbotlib.vision.VisionPipeline

Bases: ABC

An abstract vision processing pipeline

__init__(source)

Pipeline initialization

Parameters:

Name	Type	Description	Default
source	Callable[[], tuple[bool, MatLike]]	Getter for the frame to process	required

run(*args, **kwargs) abstractmethod

Runs the vision pipeline

Returns:

Type	Description
tuple[bool, MatLike | None]	tuple[bool, MatLike | None]: An OpenCV Mat for pipeline visualization purposes. Can be ignored depending on the use case.

return_values()

Retrieves the calculations from the latest pipeline iteration

Returns:

Name	Type	Description
Any	Any	Pipeline calculations

kevinbotlib.vision.EmptyPipeline

Bases: VisionPipeline

A fake vision pipeline returning the original frame

run()

Fake pipeline. Return the inputs.

Returns: Source values