Lecture 1 - Intro to Computer Vision Flashcards
What is Computer Vision?
Computers being able derive information from images, videos and other inputs. The automatic understanding of images and videos.
What can human vision do and why is it not perfect?
Recognise people and objects
Navigate through obstacles
Understand mood in the scene
Imagine stories
==
It is still not perfect though:
- Suffers from illusions
- Ignores many details
- Ambiguous description of the world
What is the understanding of measurement vs perception?
In computer vision, “measurement” refers to the process of extracting precise numerical values from an image, like the length of an object or the distance between two points, while “perception” involves interpreting and understanding the meaning of visual data, including identifying objects, recognizing patterns, and determining spatial relationships within a scene, going beyond just basic measurements
What are some factors that affect measurement vs perception?
Brightness - the lighting and shadow casted that cause objects to look different in shape, size or colour
Length Illusion - how the brain computes angles that are closer or futher
What is not Computer Vision?
Image Processing
Pattern Recognition
Computer Grapics
How did Computer Vision start?
In 1966, Minsky hired a first year undergraduate and assigned him a problem to solve: connect a television camera to a computer and get a computer to describe what it see.
What is the brief history of Computer Vision?
REFER TO SLIDES
What is an ill-posed problem?
Is more unknowns than more knowns/ number of equations
Why does Computer Vision matter - What does it help with?
Help with safety, health, security, comfort, fun and access
What are the components of a computer vision system?
Computer, camera, lightning and scene - REFER TO SLIDES
What are the challanges of Computer Vision?
Viewpoint Variation
Background Clutter
Illumination
Occlusions
Deformation
Intraclass Variations
Interclass Variations
Define Viewpoint Variation
The angle at which the video/photo is taken, the variation changes between the angle or position of the camera
Define Background Clutter
Objects or scenery in the background that can distract from the main focal point - a lot of object in the image
Define Illumination
The lighting on an object, more specifically the angle of lighting on an object
Define Occlusions
Objects that are hidden or obstructed by other objects in images or videos
Define Deformation
Change in image, typically the altering of the shape and size on an image
Define Intraclass Variations
Difference between the same object, ie species of cats
Define Interclass Variations
Difference between different objects
What are the 4 Computer Vision tasks?
Reorganisation - Segement pixels with similar properties
Reconstruction - Recover 3D information from data
Recognition - Detect and identify objects
Understanding - What is happening in the scene?
What are low level Computer Vision tasks?
Low Level Vision:
Measurements
Enhancements
Region segmentation
Features
What are some examples of low level vision?
Image Enhancement - not the original but makes the image more pleasant
Image Restoration - keeps the image close to the original
Image Inpainting - a form of image restoration
Image Compression - The process of reducing the file size of a digital image by using algorithms to remove redundant data, allowing for more efficient storage and transmission, while attempting to maintain an acceptable level of visual quality.
Super Resolution - A technique that enhances the resolution of a low-resolution image by generating missing details
Image Segmentation - Separating objects into categories, important for autonomous robotics such as self driving cars
Features - something that makes the image unique or distinct
What are mid level Computer Vision tasks?
Mid Level Vision:
Reconstruction
Depth
Motion Estimates
What are some examples of mid level computer vision?
3D Scanning
3D Maps
Depth Estimation - The process of calculating the distance between the camera and object in an image
What are high level Computer Vision tasks?
High Level Vision:
Category detection
Activity recognition
Deep understandings
Pose estimation
What are some examples of high level computer vision?
Scene Completion - Completing a image using millions of photographs to find on that is similar, using something called the nearest neighbour method
Object Detection/Recognition
Face Detection
Smile Detection
Content-Based Image Retrieval - Content-Based Image Retrieval (CBIR) is a way of retrieving images from a database. In CBIR, a user specifies a query image and gets the images in the database based on that queried image
Pose Estimation
Biometrics - fingerprint scans
Vision-based Biometrics - eye scans, xbox kinect47
What are some applications of Computer Vision?
Optical Character Recognition (OCR): Technology that converts images of text to text
Google Maps: Annotate houses/streets
Automated visual inspections
Object recognition - Amazon Go
Vision-based biometrics - Apple face id, fingerprint scanners
Human shape captures
Motion capture
Medical imaging
Virtually guided surgery
Virtual Reality
Augmented Reality
Body Tracking
Smart Cars
Mobile Robots
Drone
Industrual Robots/Robotics
Human Detection
Video Surveillance and Monitoring
Forest Fire Monitoring System
Counting in Crowded Images
Fatigue Detection
Lip- reading
Human Recognition
Morphing
What are some applications of Computer Vision - specifically in technology?
❖ Laptop: Biometrics auto-login (face recognition, 3D), OCR
❖ Smartphones: QR codes, computational photography (Android Lens Blur, iPhone Portrait Mode), panorama construction (Google Photo Spheres), face detection,
expression detection (smile), Snapchat filters (face tracking), Google Tango (3D reconstruction), Night Sight (Pixel)
❖ Web: Image search, Google photos (face recognition, object recognition, scene recognition, geolocalization from vision), Facebook (image captioning), Google maps aerial imaging (image stitching), YouTube (content categorization)
❖ VR/AR: Outside-in tracking (HTC VIVE), inside out tracking (simultaneous localization and mapping, HoloLens), object occlusion (dense depth estimation)
❖ Motion: Kinect, full body tracking of skeleton, gesture recognition, virtual try-on
❖ Medical imaging: CAT / MRI reconstruction, assisted diagnosis, automatic pathology, connectomics, endoscopic surgery
❖ Industry: Vision-based robotics (marker-based), machineassisted router (jig), automated post, ANPR (number plates), surveillance, drones, shopping
❖ Transportation: Assisted driving (everything), face tracking/iris dilation for drunkeness, drowsiness, automated distribution
❖ Media: Visual effects for film, TV (reconstruction), virtual sports replay (reconstruction), semantics-based auto edits
