demo Namespace Reference

Functions
	setup_model ()
	Initializes and load the parameters of the MSE-CNN.
	int2label (split)
	Obtain the string that corresponds to an integer value of the split.
	draw_partition (img, split, cu_pos, cu_size)
	Draw partition in image based in the split outputed by the model.
	split_fm (cu, cu_pos, split)
	Splits feature maps in specific way.
	partition_img (img, img_yuv)
	Partitions a full 128x128 CTU and draws the partition in the original image.
	pipeline (img, text)
	Pipeline to implement the functionalities to demonstrate the potential of the MSE-CNN.
	main ()

Variables
str	PATH_TO_COEFFS = "../../../model_coefficients/best_coefficients"
str	LOAD_IMAGE_ERROR = "load_image_error.png"
list	EXAMPLE_IMGS = ["example_img_1.jpeg", "example_img_2.jpeg"]
tuple	CTU_SIZE = (128, 128)
	FIRST_CU_POS = torch.tensor([0, 0]).reshape(shape=(-1, 2))
	FIRST_CU_SIZE = torch.tensor([64, 64]).reshape(shape=(-1, 2))
str	DEV = "cuda" if torch.cuda.is_available() else "cpu"
int	QP = 32
	model = None
tuple	COLOR = (0, 247, 255)
int	LINE_THICKNESS = 1
str	DEFAULT_TEXT_FOR_COORDS = "Insert CTU position in the image..."

Detailed Description

@package docstring 

@file demo.py 

@brief Demonstration of the application of the MSE-CNN 

Note: In order to run this script, you have to do it inside the folder

@section libraries_demo Libraries 
- msecnn
- train_model_utils
- cv2
- dataset_utils
- re
- sys
- numpy
- gradio
- torch
- custom_dataset
- PIL
 
@section classes_demo Classes 
- None 

@section functions_demo Functions 
- setup_model()
- int2label(split)
- draw_partition(img, split, cu_pos, cu_size)
- split_fm(cu, cu_pos, split)
- partition_img(img, img_yuv)
- pipeline(img, text)
- main()

@section global_vars_demo Global Variables 
- PATH_TO_COEFFS
- LOAD_IMAGE_ERROR
- EXAMPLE_IMGS
- CTU_SIZE
- FIRST_CU_POS
- FIRST_CU_SIZE
- DEV
- QP
- model
- COLOR
- LINE_THICKNESS
- DEFAULT_TEXT_FOR_COORDS

@section todo_demo TODO 
- Instead of obtaining the best split, do the thresholding and then split it until you find the right type of split 

@section license License 
MIT License 
Copyright (c) 2022 Raul Kevin do Espirito Santo Viana
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

@section author_demo Author(s)
- Created by Raul Kevin Viana
- Last time modified is 2023-09-10 21:00:10.225508

Function Documentation

draw_partition()

demo.draw_partition	(		img,
			split,
			cu_pos,
			cu_size
	)

Draw partition in image based in the split outputed by the model.

Parameters

[in]	img	User's input image
[in]	cu_pos	CU position
[in]	cu_size	CU size
[in]	split	Integer number representing the split that the model chose
[out]	str_split	Name of the corresponding split

int2label()

demo.int2label

(

split

)

Obtain the string that corresponds to an integer value of the split.

Parameters

[in]	split	Integer number representing the split tht the model chose
[out]	str_split	Name of the corresponding split

main()

demo.main

(

)

partition_img()

demo.partition_img	(		img,
			img_yuv
	)

Partitions a full 128x128 CTU and draws the partition in the original image.

TODO: Instead of obtaining the best split, do the thresholding and then split it until you find the right type of split

Parameters

[in]	img	Image in BGR
[in]	img_yuv	Image in YUV
[in]	stg	Current stage being partitioned
[in]	cu_pos	Current stage being partitioned
[in]	cu_size	Current stage being partitioned
[out]	img	Image in with partitions drawn to it

pipeline()

demo.pipeline	(		img,
			text
	)

Pipeline to implement the functionalities to demonstrate the potential of the MSE-CNN.

Parameters

[in]	img	Image in RGB
[out]	mod_img	Modified image with drawings into it in RGB
[out]	best_split	Best split (BTV, BTH, TTV, TTH, Non-split, QT)

setup_model()

demo.setup_model

(

)

Initializes and load the parameters of the MSE-CNN.

split_fm()

demo.split_fm	(		cu,
			cu_pos,
			split
	)

Splits feature maps in specific way.

Parameters

[in]	cu	Input to the model
[in]	cu_pos	Coordinate of the CU
[in]	split	Way to split CU
[out]	cu_out	New Feature maps
[out]	cu_pos_out	Position of the new CUs

Variable Documentation

COLOR

tuple demo.COLOR = (0, 247, 255)

CTU_SIZE

tuple demo.CTU_SIZE = (128, 128)

DEFAULT_TEXT_FOR_COORDS

str demo.DEFAULT_TEXT_FOR_COORDS = "Insert CTU position in the image..."

DEV

str demo.DEV = "cuda" if torch.cuda.is_available() else "cpu"

EXAMPLE_IMGS

list demo.EXAMPLE_IMGS = ["example_img_1.jpeg", "example_img_2.jpeg"]

FIRST_CU_POS

demo.FIRST_CU_POS = torch.tensor([0, 0]).reshape(shape=(-1, 2))

FIRST_CU_SIZE

demo.FIRST_CU_SIZE = torch.tensor([64, 64]).reshape(shape=(-1, 2))

LINE_THICKNESS

int demo.LINE_THICKNESS = 1

LOAD_IMAGE_ERROR

str demo.LOAD_IMAGE_ERROR = "load_image_error.png"

model

demo.model = None

PATH_TO_COEFFS

str demo.PATH_TO_COEFFS = "../../../model_coefficients/best_coefficients"

QP

int demo.QP = 32