3-day Virtual Seminar

Image Processing and Computer Vision: From Pixels to Intelligent Visual Recognition

Overview

This webinar provides an accessible introduction to Image Processing and Computer Vision, beginning with the representation of images and progressing toward object recognition and modern deep-learning-based vision systems.

The session first introduces spatial filtering, including Gaussian and median filtering, and explains how filtering can be used for smoothing, noise suppression, and preservation of important image structures.

The webinar then introduces edge detection, explaining gradients, gradient magnitude and direction, and the importance of non-maximum suppression in obtaining thin and accurately localized edges.

The session subsequently introduces the image formation model and camera calibration, including intrinsic and extrinsic parameters, coordinate systems, projection, lens distortion, and Zhang's camera calibration approach.

The next stage focuses on local feature detection and description. Scale-space and Difference of Gaussian are used to understand how keypoints can be detected at characteristic scales. SIFT is then introduced as a scale- and rotation-invariant feature descriptor. The webinar explains canonical orientation, local coordinate frames, orientation histograms, descriptor construction, and feature matching.

The discussion then moves to instance-level object detection, covering template matching, shape-based matching, Hough Transform, Generalized Hough Transform, and model-based feature matching. These methods demonstrate how a known object can be located in a new image even when its position, scale, or orientation changes.

Finally, the webinar connects classical Computer Vision to Deep Learning and CNN architectures, introducing the evolution from LeNet to AlexNet, ZFNet, VGG, GoogleNet/Inception, and ResNet. Important architectural ideas such as pooling, convolutional layers, grouped convolutions, bottleneck blocks, skip connections, global average pooling, and computational efficiency are discussed.

The overall goal is to provide participants with a coherent understanding of how a computer progresses from raw pixels to meaningful visual information and object recognition.

Why you should Attend

AI systems can recognize faces, objects, scenes, and visual patterns-but what actually happens between an image entering the computer and an object being recognized?

Understanding Computer Vision requires more than knowing that "CNNs recognize images." The underlying ideas begin much earlier: filtering removes unwanted information, gradients reveal changes, edges identify boundaries, local features provide distinctive points, descriptors represent local appearance, and matching establishes correspondences between images.

This webinar will help participants understand the complete evolution of visual recognition, from classical image-processing techniques to modern deep-learning architectures.

Participants will also understand why techniques such as SIFT, Hough Transform, and camera calibration remain important for understanding the principles behind modern computer vision.

Areas Covered in the Session

• Introduction to images and image processing 
  
• Spatial filtering and noise reduction 
  
• Edge detection and Non-Maximum Suppression 
  
• Image formation and camera calibration 
  
• Scale-space and Difference of Gaussian (DoG) 
  
• SIFT keypoints and feature descriptors 
  
• Feature matching and Lowe's ratio test 
  
• Template Matching and Hough Transform 
  
• Generalized Hough Transform and instance detection 
  
• CNNs: LeNet, AlexNet, VGG, Inception and ResNet 
  
• Convolution, pooling, Inception modules and residual connections 
  
• Evolution from classical Computer Vision to Deep Learning
  

Who Will Benefit

• Computer Science Students
  
• Engineering Students
  
• Artificial Intelligence / Machine Learning Students
  
• Computer Vision Students
  
• Robotics Students
  
• Data Science Students
  
• Software Developers interested in AI and Computer Vision
  
• Aspiring AI/ML Engineers
  
• Students studying Image Processing
  
• Students studying Pattern Recognition
  
• Researchers interested in Computer Vision
  
• Students preparing for Computer Vision and AI examinations
  
• Anyone interested in how machines understand images