Udemy
    •  
    •  
    •  
    •  
    •  
    •  
    •  
    •  
Subscribe
Turn what you know into an opportunity and reach millions around the world.
Learn More
Your cart is empty.
Keep shopping
Python/Django App- Create & Deploy a Computer Vision Model
Rating: 4.3 out of 5(75 ratings)
468 students

Python/Django App- Create & Deploy a Computer Vision Model

Full Stack Computer Vision web app using python and Django, Transfer Learning, CNN, Keras, html, CSS, JavaScript, Ajax.
Created byAshar Siddiqui
Last updated 5/2021
English

What you'll learn

  • Creating a full stack computer vision model using Transfer Learning in Python. The course will include details on how to create a computer vision model in python, and how to host it on server using Django.
  • How to save and deploy any python ML/DL model you have created using Django.
  • How to deploy a model in Production, Client Side(html, CSS) and Server side(Python) programming. All open source and free to use technologies.
  • Learn Django and Integrating a python code with the Django Framework.
  • How to create a user interface(UI) for your python code or ML/DL model that can take input from user, pass the input to your ML/DL model and renders back the results to UI.
  • How to utilize transfer learning for feature extraction thus helping train new models without the need of a powerful GPU.
  • Re-usability : how to quickly retrain the model that you create on new set of images.
  • How to create an end to end computer vision project.

Course content

8 sections • 43 lectures • 7h 20m total length

  • Course Structure and Contents4:23
  • Proof Of Concept - Car Damage Detection7:29

    This is a proof of concept created for possible automation of the car damage detection process. The model has been created using CNN Architecture on Keras and transfer learning utilizing the VGG16 model. Django 1.10 has been used to create a full stack website.

  • POC 2.0 - Single page portal without refresh using AJAX6:17

    This POC has been created using AJAX. The intent is to render the results on the same page without refresh. Single page portal means you can upload the image and get the results on the same page. The home page for the portal is still there and it can be utilized if you intent to add more features. If you want, you can remove it and can create a portal with only one page.

  • POC 3.0 - Integrating KYC functionality to the portal7:51

    POC 3.0 will have KYC functionality integrated which will enable user to identify the type of id card uploaded, based on training. The classifier used in this POC has been trained on Pan card and Adhaar card.

  • Upgrading to the latest Django Version0:31

    Important information about Django Version

    At the time of developing this app, I utilised the Django version 1.10.

    My recommendation would be to follow along the course and create the POC using Django version 1.10. Once you have successfully created the application, then you can go ahead and upgrade to the latest Django version. As of May 2021, the latest Django version is 3.2.2

    Try to upgrade the Django version on your own. Its pretty straight forward and very simple upgrade. In case of any challenges, I have uploaded a new section(Section 8) at the bottom of this course. It contains all the code files, models and templates etc along with requirements.txt file that you can use to upgrade the application.

    Thanks and wish you all the best. Happy learnings!!!

  • Installation - Anaconda, Django and Atom5:58
  • Anaconda Prompt Basics8:15

    Learn how to create  and activate/deactivate Anaconda Environment, how to launch Jupyter Notebook from Anaconda prompt, basics of python and how to write functions and class in Jupyter Notebook.

  • Working with Jupyter Notebook8:57

Requirements

  • Python (I did included some python basics as a refresher)
  • Knowledge of Deep learning good to have but not a must. I did included detailed lectures explaining CNN architecture and transfer learning, along with code.
  • Zeal to learn.

Description

This Course has been designed for the developers who are able to train ML/DL models, but they struggle when it comes to saving the model for future use or when it comes to deploying the model through a full stack portal.

This course will teach you how to train and create computer vision model from scratch, how to utilize transfer learning for feature extraction, how to save those models using pickle,  and how to deploy the models using Django framework.

Who this course is for:

  • One who wants to create full stack portal with client side(html, css, javascript) and server side(Python) functionality.
  • One who wants to save his trained ML/DL model in python for future predictions.
  • One who knows how to create a ML/DL model in python but don't know how to deploy it.
  • One who wants to host his model as Web Server.
  • Students who want to create a project. The models can be retrained on new set image really quickly and projects like KYC or any other image classification projects can be created end to end.
  • One who wants to code practical implementation using open source libraries like tensorflow and Keras.

Report abuse