You are on page 1of 9

COMPUTER VISION

Done By Michael
Computer vision pipeline
• A Computer Vision Pipeline is a series of steps that
most computer vision applications will go through.
Many vision applications start by acquiring images
and data, then processing that data, performing
some analysis and recognition steps, then finally
performing an action.
General pipeline of CV
Image Classification
 It’s the process of extracting information from the images and labelling or categorizing
the images
 For example in this case study Rethinking the Inception Architecture for Computer
Vision will be using Convolutional Neural Networks.
 CNN is a neural network which helps us extract important features from an image that
will help us provide learnable parameters to efficiently do the classification.
 CNN is a deep learning algorithm which takes image as an input.
 In this scenario Inception v3 which is 48 layers deep will be out convolutional neural
network
 Also can be called GoogleNetv3
Pre-Processing
• It is preliminary processing of data in order to
prepare it for primary processing or further
analysis.
• It is done in order to improve the image data
that suppresses unwilling distortions or
enhances some image features important for
further processing
Pre processing cont.

o There are four different types of pre-processing which are:

1. Pixel brightness transformations


2. Geometric Transformations
3. Image Filtering and Segmentation
4. Fourier Transform and Image restauration.
Key Words to note in Feature extraction
• Kernel or filter (k)- it is also called feature detector used to
extract the important features from the image. It is a matrix
that moves over input data, preforms the dot product with the
sub region of input data and gets the output as the matrix
• Padding- Border problem is solved
• Max pooling
• Average pooling – it reduce the dimension without loosing
important features.
• Flattening involves transforming the entire pooled feature map
matrix into a single column which is then fed to the neural
network (CNN) for processing.
Bonuses
• Improving Accuracy of CNN
Use bigger pretrained models
Use k-Fold cross optimization
Use MixUp to augment your images
Applications of Image classification

• Filters on Editing Apps and social Media


• Medical Technology (Detection of cancer cells in
pathology cells)
• Pattern recognition (Face recognition in security)
• Traffic monitoring and congestion detection

You might also like