Introduction

Computer vision is an interesting area of Machine Learning and Deep Learning that still needs to evolve a lot to match human-level accuracy. So, I believe to inorder to increase the community strength for computer vision, we need more people equipped to work in this field and hence such a session on computer vision will be important. Hence this session will be a hands-on workshop on the introductory concepts of computer vision using deep learning algorithms and concepts.

Objective:

The objective of this workshop will be to teach the audience on the following concepts of Computer Vision:

  1. Image Manipulations like applying filters to an image
  2. Convolutional Neural Network
  3. Different CNN Architectures
  4. Image Classification
  5. Data Augmentation and other regularization techniques
  6. Transfer Learning
  7. Object Detection
  8. Neural Style Transfer
  9. Image Generation

The target will be to teach the audience with introductory to intermediate level of knowledge on the above topics. Based on the audience background and interest, the level of details can be modified.

Programming language will be Python. Preferred framework can be keras, tensorflow and open cv. Preferred medium of training will be Anaconda Spyder IDE or Jupyter notebook.

The idea is to teach the fundamental concepts in each area, show sample code-snippet for each applications and provide equivalent assignments to the audience, so that the audience can learn by hands-on coding and gain practical experience on this area.

 
1 favorite thumb_down thumb_up 0 comments visibility_off  Remove from Watchlist visibility  Add to Watchlist
 

Outline/Structure of the Workshop

The rough structure of the workshop will look like:

  1. Introduction to Computer Vision using Deep Learning
    1. Brief discussion on Computer Vision with Classical Machine Learning approach
    2. Fundamental concepts of computer vision
  2. Image Manipulations like applying filters to an image
    1. Applying pencil sketch filters in images
    2. Producing transparent images
    3. Applying heat maps on images
  3. Convolutional Neural Network
    1. Introduction to Convolutional Neural Networks
    2. Discussions on CNN concepts like padding, pooling, strided convolution etc.
    3. Importance of filters and different types of filters
    4. Edge detection using horizontal and vertical filters
    5. Matrix dimensionality during CNN operation
  4. Different CNN Architectures
    1. Brief discussion on LeNet-5
    2. Brief discussion on AlexNet
    3. Brief discussion on VGG-16
    4. Brief discussion on ResNet
    5. Brief discussion on InceptionNet
  5. Image Classification
    1. Image Classification using CNN and Logistic Regression
    2. Image Classification using CNN and Softmax activation
  6. Data Augmentation and other regularization techniques (with practical example and use cases)
  7. Transfer Learning (with practical example and use cases)
  8. Object Detection
    1. Object detection with YOLO algorithm
    2. Objection using R-CNN
  9. Neural Style Transfer
  10. Image Generation
    1. Using Generative Adversarial Networks
    2. Using Variational Auto Encoders

I am planning to give small problem statements/assignments/challenges which the audiences have to complete on each of these topics along with discussion/lecture/practical demo, so that the audience can receive hands-on experience

Learning Outcome

  1. Learn and understand popular concepts and topics related to computer vision using deep learning
  2. Hands-on experience on computer vision using Python and popular frameworks
  3. Gain experience in handling image data
  4. Gain experience in working with interactive problem statements and challenges or mini projects
  5. Learn and work on some of the latest algorithms on Computer Vision with deep learning

Target Audience

Data Scientists, Engineers, Developers, AI Enthusiasts, particularly Computer Vision enthusiasts

Prerequisites for Attendees

  1. Basic knowledge on Machine Learning
  2. Basic knowledge in Python programming
  3. Previous knowledge of working with Jupyter notebook or Spyder IDE
  4. Basic knowledge of Neural Networks
schedule Submitted 1 week ago

Public Feedback

comment Suggest improvements to the Speaker