MultiVisionToolkit

MultiVisionToolkit is a Python package that provides tools for object detection and segmentation, specifically using the YOLOv8 model. It includes real-time detection on camera video, visualization metrics, and the ability to convert reports to document and PDF files.

Installation

pip install multivision

Usage

Object Detection with YOLOv8

Detailed feature showcase with images:

Download images for created dataset

#download images form internet
#using class name and count for images more than 500 images for best training 
from multivision.dataset import dataset as ds
class_name='dog'
count=100
ds.download_images(class_name,count)

Extract imges from any video

from multivision.dataset import dataset as ds
video_path="video.mp4"
images_folder_path="images"
frame_strid=10
ds.extract_images_video(video_path,images_folder_path,frame_strid)

download video from youtube and extract to images

from multivision.dataset import dataset as ds
video_url="https://www.youtube.com/shorts/6eb9-P6KHN0"

#ds.download_yt(video_url, output_path='.')
#or
output_path="images"
ds.download_yt(video_url, output_path)

Annotation auto label for dataset without any manual tools

from multivision.annotation import autolabel as auto
ontology_dict=auto.create_ontology_dict() #create caption for custom dataset
image_folder=path_of_images_folder"
dataset_folder="dataset_folder_to_save_train,val with images labels "
auto.create_captions(ontology_dict, image_folder, dataset_folder)

training custom datatset with yaml data file for detection training

from multivision.train import yolov8 as y8
model_det_name="yolov8n.pt"
epochs_no=5
data_yaml_path="E:/multivision/dataset/data.yaml"
y8.y8d_train(model_det_name,epochs_no,data_yaml_path)

training custom datatset with yaml data file for segmentation training

model_seg_name="yolov8n-seg.pt"
epochs_no=10
y8.y8s_train(model_name,epochs_no,data_yaml_path)

Visualization Metrics

from multivsion.visualize import vis as vis
vis.images_google_colab(folder_path)
vis.display_images_cv(folder_path,scale_factor)
vis.display_images_with_grid(folder_path, rows, cols)
vis.plot(annotation_path,images_dir_path,yaml_path,samples_no)

Convert Report to Document and PDF

# Example Usage:
folder_path = "images"
output_docx = "output_document.docx"
title = "My Document Title"
author = "Your Name"
pdf_path="book.pdf"
copyright_notice = "© 2023 Falah.G.Salieh"

conclusion = "This is the conclusion of the document."
from multivision.docx import document as doc
from multivision.docx import document as pdf
doc.create_word_document(folder_path, output_docx, title, author, conclusion)
pdf.images_to_pdf(folder_path, pdf_path, title, copyright_notice)

License

This project is licensed under the MIT License - see the LICENSE file for details.

Citation

If you find MultiVisionToolkit helpful in your work, please consider citing it. You can use the following BibTeX entry:

@software{multivisiontoolkit,
  author = {Falah.G.Salieh},
  title = {MultiVisionToolkit},
  year = {2023},
  url = {https://github.com/falahgs/multivisiontoolkit},
}