Please enable JavaScript.
Coggle requires JavaScript to display documents.
Object Recognition (The Theory), Muhammad Asyraaf Syahmi bin Rosmawee β¦
Object Recognition (The Theory)
Brief History
1966, Minky assigns computer vision
Late 1960, Interpretation of synthentic words
1970, some progress on interpreting selected images
1980, ANN come and go. Shift toward geometry and increased mathematical rigor
1990, face recognition, statistical analysis in vogure
2000, broader recognition, large annotated datasets available, video processing starts
2030, robot uprising
Recognition problem is difficult with computer
Human being easy to recognize objects by color, texture and appearance
A spontaneous activity, natural activity for humans and other biological systems.
But in case of computers, they are not able to recognize just with appearance
Problem faced by computer while detecting objects
Scale and shape of the imaged object varies with viewpoint
Occlusion (self or by a foreground object)
Lighting variations
Background "clutter"
Human Vision
Vision is the process of discovering what is present in the world and where it is
Scene -> Eye -> Image -> Brain -> Perception
Computational algorithm implemented in this massive network of neurons
Computer Vision
Study of analysis of pictures and videos in order to achieve results like those as by humans
Scene -> Camera -> Image -> Computer -> Perception
Main goal
Every picture/image tells a story
Write a computer programs that can interpret images
Signifcance of Computer Vision
Safety
Healthcare
Security
Convenient
Entertainment
Access
Digital Image Processing
Low Level Process
Noise removal, image sharpening
Mid Level Process
Object recognition, segmentation
High Level Process
Scene understanding, autonomous navigation
Object recognition
One of the major domain in image processing
Task of finding and identifying objects in an image or video sequence
Goal of instance-level recognition is to match (recognize) a specific object or scene
Human understanding
Detection of separate objects
Description of their geometry and position in 3D
Classification as being one of a known class
Identification of the instance/occurence
Understanding of spatial relationships between objects
Starting from an image of an object of interest (the query), search through an image dataset to obtain (or retrieve) those images that contain the target obejct
Learning and Adaption
Supervised learning
Unsupervised learning
Semi-supervised learning
Pattern Recognition
Assignment of a physical object or event to one of several prespecified categories
A pattern is an object, process or event that can be given a name/ label/ mark
A pattern class or category is a set of patterns sharing a common attributes and usually originating from same source
During recognition(classification) given objects are assigned to prescribed classes
A classifier is a machine which performs classification
Components of a pattern recognition system
Data acquisition and sensing
Pre-processing
Feature extraction
Model learning and estimation
Classification
Post-processing
Feature Extraction
To extract features which are good for classification
Supervised methods
Unsupervised methods
Classifier
π = π1 βͺ π2 βͺβ¦ βͺ π|π| and π1 β© π2 β©β¦ β© π|π| = 0
Consist of determining to which region a feature vector X 'belongs to. Borders between decision boundaries are called decision region
Feature
Feature is a scalar x which is quantitatively describes a property of the object
Feature extraction consist of choosing those features which are most effective for preserving class responsibility
Pattern
n-tuple X (vector) of N scalars xi for i β [1, N], which are called the features
Class
A set of patterns that share some common properties
Classification is a mathematical function or algorithm which assigns a feature to one of the classes
Cluster separation
Misclassifications are a consequence of the separation of the clusters. The separation of clusters is quantified using two major methods
Mathematically, there are several separation criteria's
Intuitively, overlapping of the clusters
Classification quality
Muhammad Asyraaf Syahmi bin Rosmawee
(AI200155)
Asynchronous 3