Scribd
Upload
5 views2 pages

Evolution of Vision: Biology to Machine

The document discusses the evolution of biological and machine vision, highlighting the development of visual systems from the Cambrian period to modern mammals. It covers key discoveries in visual processing by Hubel and Wiesel, as well as advancements in machine vision through deep learning and traditional methods. Notable models such as the Neocognitron, LeNet-5, and AlexNet are mentioned as significant contributions to the field.

Uploaded by

Anitha p c Anitha p c
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
5 views2 pages

Evolution of Vision: Biology to Machine

The document discusses the evolution of biological and machine vision, highlighting the development of visual systems from the Cambrian period to modern mammals. It covers key discoveries in visual processing by Hubel and Wiesel, as well as advancements in machine vision through deep learning and traditional methods. Notable models such as the Neocognitron, LeNet-5, and AlexNet are mentioned as significant contributions to the field.

Uploaded by

Anitha p c Anitha p c
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

MODULE 1

Biological and Machine Vision

Biological Vision

During the prehistoric Cambrian period (about 550 million years ago), the number of species on
Earth increased rapidly.

This explosion was driven by the development of light detectors in trilobites.

Vision enabled organisms to detect food, predators, and mates with high accuracy and speed.

This triggered an evolutionary arms race, leading to the Cambrian explosion.

Evolution of Visual Complexity

Over millions of years, visual systems became more complex.

In modern mammals, a large portion of the cerebral cortex is dedicated to visual perception.

Gray matter performs complex computation, while white matter transmits information rapidly.

Hubel and Wiesel’s Discoveries

Hubel and Wiesel studied how visual information is processed in the primary visual cortex.

They discovered simple cells that respond to edges at specific orientations.

Complex cells combine inputs from simple cells to detect complex shapes.

Hierarchical Visual Processing

Visual information is processed through multiple layers.

Early layers detect edges and lines.

Deeper layers detect complex shapes and objects.

Specific regions of the brain detect color, motion, and faces.

Machine Vision

Machine vision is inspired by biological vision.

There are two approaches: deep learning and traditional machine learning.

The Neocognitron

Proposed by Kunihiko Fukushima.

Inspired by biological simple and complex cells.


Uses hierarchical artificial neurons to detect features.

LeNet-5

Developed by Yann LeCun and Yoshua Bengio.

First successful convolutional neural network.

Uses backpropagation for learning.

Used by the US Postal Service for ZIP code recognition.

Traditional Machine Learning Approach

Relies heavily on feature engineering.

Example: Viola–Jones face detection algorithm.

Uses handcrafted features for face detection.

ImageNet and ILSVRC

ImageNet is a large labeled image dataset created by Fei-Fei Li.

ILSVRC became a benchmark for machine vision.

AlexNet

Developed in 2012 by Krizhevsky, Sutskever, and Hinton.

Marked a major breakthrough in deep learning.

Enabled by GPUs, large datasets, and improved architecture.

TensorFlow Playground

Interactive tool to visualize deep learning.

Demonstrates hierarchical feature learning.

Summary

Deep learning is inspired by biological vision.

Hierarchical models learn complex features automatically.

AlexNet revived deep learning in machine vision.

You might also like