Computer Perception Extended

WEEK 04

* * *

Main repo

Status Quo?

1. Convolutional Neural Networks

1.1 Concept

Examples

Examples

The LeNet Architecture (1990s)

Local Patterns

Abstract visual concepts

Feature map

1.2 Steps

Neural Network

Input

Brute Force Idea 1: Searching with a Sliding Window

Brute Force Idea 2: More data and a Deep Neural Net

Step 1: Break the image into overlapping image tiles

Step 2: Feed each image tile into a small neural network

Step 3: Save the results from each tile into a new array

Step 4: Downsampling

Step 4: Downsampling

Final step: Make a prediction

1.3 How Convolution Works

The LeNet Architecture (1990s)

Convolution of the 5 x 5 image and the 3 x 3 matrix

Convolution Demo 01

Convolution Demo 02

Parameters

Pooling

Summary

Visualising CNN

Understanding Neural Networks Through Visualization

Credits