ARTIFICIAL VISION

International Teaching ARTIFICIAL VISION

Back

0622700045
DEPARTMENT OF INFORMATION AND ELECTRICAL ENGINEERING AND APPLIED MATHEMATICS
EQF7
COMPUTER ENGINEERING
2024/2025

OBBLIGATORIO
YEAR OF COURSE 2
YEAR OF DIDACTIC SYSTEM 2022
AUTUMN SEMESTER
CFUHOURSACTIVITY
432LESSONS
18LAB
18EXERCISES
Objectives
THE COURSE AIMS AT PROVIDING THE COMPETENCES ON THE MAIN METHODOLOGIES AND TECHNIQUES REQUIRED TO REALIZE AN ARTIFICIAL VISION SYSTEM.

KNOWLEDGE AND UNDERSTANDING
KNOWLEDGE OF THE DIFFERENT TASKS CARRIED OUT WITHIN AN ARTIFICIAL VISION SYSTEM, AND IN PARTICULAR WITH REGARDS TO THE LOW LEVEL PROCESSING PHASES (ACQUISITION, FILTERING), TO THE INTERMEDIATE LEVEL PHASES (REGIONALIZATION AND CONTOURS EXTRACTION) AND TO THE HIGH LEVEL PROCESSING (SHAPE RECOGNITION, TRACKING), AS WELL AS UNDERSTANDING OF THE BASIC TECHNIQUES FOR IMPLEMENTING SUCH FUNCTIONS.

APPLYING KNOWLEDGE AND UNDERSTANDING
DIMENSION AN IMAGE AND / OR A VIDEO CAPTURE SYSTEM SATISFYING THE REQUIREMENTS. DESIGN AND IMPLEMENT AN ARTIFICIAL VISION SYSTEM FOR THE INTERPRETATION OF IMAGES AND / OR VIDEOS USING FUNCTIONS OF THE OPENCV COMPUTER VISION SOFTWARE LIBRARY INTEGRATING THEM WITH MACHINE LEARNING TECHNIQUES.

Prerequisites
IN ORDER TO ACHIEVE THE GOALS OF THE COURSE, IT IS REQUIRED THE KNOWLEDGE OF THE PYTHON PROGRAMMING LANGUAGE AND OF THE MAIN MACHINE LEARNING AND DEEP LEARNING FRAMEWORKS AS KERAS AND TENSORFLOW.
Contents
Didactic unit 1: Architecture of an artificial vision system
(LECTURE/PRACTICE/LABORATORY HOURS 4/0/0)
- 1 (2 Hours Lecture)): Course introduction. HISTORICAL INTRODUCTION TO THE ARTIFICIAL VISION SYSTEMS.
- 2 (2 Hours Lecture)): Architecture of an artificial vision systems and its processing phases
KNOWLEDGE AND UNDERSTANDING: General architecture of an artificial vision system
APPLYING KNOWLEDGE AND UNDERSTANDING: Recognize the different components and processing stages of an artificial vision system.

Didactic unit 2: Image acquisition, representation and preprocessing
(LECTURE/PRACTICE/LABORATORY HOURS 10/6/0)
- 3 (2 Hours Lecture): Architecture of an image acquisition system. Optics: pinhole and thin lens models. Key concepts: focal length, angle of view, depth of field, aperture, exposure.
- 4 (2 Hours Lecture): Sensor types: visible, infrared, thermal. Elements of depth imaging.
- 5 (2 Hours Practice): Methods and tools for the design of a camera acquisition system in a real scenario. Discussion of an example
- 6 (2 Hours Lecture): Image representation. Color spaces: RGB and HSV. Color temperature.
- 7 (2 Hours Practice): Introduction to OpenCV. Basic operation with images: load from/save to file, image/video capture and display
- 8 (2 Hours Lecture): Thresholding. Smoothing filters. Derivative filters. Morphological operators.
- 9 (2 Hours Lecture): Canny edge detection algorithm. Connected component labeling.
- 10 (2 Hours Practice): Presentation and discussion of a pipeline based on low level and intermediate level processing stages for facing a real an artificial vision problem.
KNOWLEDGE AND UNDERSTANDING: Main parameters of an image acquisition system. Main image preprocessing techniques, fields of application, advantages and limitations. OpenCV software library for computer vision.
APPLYING KNOWLEDGE AND UNDERSTANDING: Identify the image acquisition system starting from the specifications relating to the distance of the objects to be analyzed, the framing area, the minimum resolution. Identify the most appropriate image pre-processing and regionalization techniques for the specific computer vision problem.

Didactic unit 3: Identification and recognition of objects
(LECTURE/PRACTICE/LABORATORY HOURS 10/4/0)
- 11 (2 HOURS Lecture): DETECTION AND RECOGNITION OF OBJECTS WITH TRADITIONAL APPROACHES: VIOLA JONES ALGORITHM
- 12 (2 HOURS Lecture): Object detection by deep learning approaches
- 13 (2 HOURS Practice): SOFTWARE IMPLEMENTATION OF AN OBJECT DETECTION SYSTEM
- 14 (2 HOURS Lecture): FACE ANALYSIS ALGORITHMS: EXTRACTION OF TRUST POINTS BY NORMALIZATION OF FACES
- 15 (2 HOURS Lecture): Principal Component Analysis for Face Recognition
- 16 (2 HOURS Lecture): DETECTION, VERIFICATION AND RECOGNITION OF FACES
- 17 (2 HOURS Practice): SOFT BIOMETRICS (CHARACTERIZATION OF AGE, SEX, Ethnicity, EMOTION)
KNOWLEDGE AND UNDERSTANDING: Object detection and recognition using traditional machine learning techniques and deep learning techniques. In-depth study of the problem of the analysis of human faces.
APPLYING KNOWLEDGE AND UNDERSTANDING: Apply traditional machine learning and deep learning techniques to build an object detection and recognition system. Use low- and mid-level techniques to prepare data for the training phases of the machine learning system. Evaluate the performance of the implemented system.

Teaching unit 4: AUTOMATIC INTERPRETATION OF VIDEO
(LECTURE/PRACTICE/LABORATORY HOURS 6/0/0)
- 18 (2 HOURS Lecture): TECHNIQUES FOR TRACKING SINGLE OBJECTS IN MOTION FROM A FIXED OR MOBILE CAMERA
- 19 (2 HOURS Lecture): TECHNIQUES FOR TRACKING MULTIPLE MOVING OBJECTS FROM A FIXED CAMERA BY SUBTRACTING AND UPDATING THE BACKGROUND
- 20 (2 HOURS Lecture): Presentation of commercial applications for computer vision
KNOWLEDGE AND UNDERSTANDING: Video sequence analysis techniques. Maintenance and updating of the background. Tracking of objects of interest from fixed and mobile cameras.
APPLYING KNOWLEDGE AND UNDERSTANDING: Design and implement an artificial vision system for the analysis of video streams using the algorithms already available in the OpenCV library.

Teaching unit 5: PROJECT WORK
(LECTURE/PRACTICE/LABORATORY HOURS 0/0/8)
- 21 (2 HOURS Laboratory): Presentation of the final project of the course
- 22 (2 HOURS Laboratory): Implementation of the final project of the course
- 23 (2 HOURS Laboratory): Implementation of the final project of the course
- 24 (2 HOURS Laboratory): Implementation of the final project of the course
KNOWLEDGE AND UNDERSTANDING: -
APPLYING KNOWLEDGE AND UNDERSTANDING: Design and implement a a complete artificial vision system using traditional machine learning and deep learning techniques.


TOTAL LECTURE/PRACTICE/LABORATORY HOURS 30/10/8
Teaching Methods
THE COURSE CONTAINS THEORETICAL LECTURES, IN-CLASS EXERCITATIONS AND PRACTICAL LABORATORY EXERCITATIONS. DURING THE IN-CLASS EXERCITATIONS THE STUDENTS ARE DIVIDED IN TEAMS AND ARE ASSIGNED SOME PROJECT-WORKS TO BE DEVELOPED ALONG THE DURATION OF THE COURSE. THE PROJECTS INCLUDE ALL THE CONTENTS OF THE COURSE AND IS ESSENTIAL BOTH FOR THE ACQUISITION OF THE RELATIVE ABILITIES AND COMPETENCES, AND FOR DEVELOPING AND REINFORCING THE ABILITY TO WORK IN A TEAM. IN THE LABORATORY EXERCITATIONS THE STUDENTS IMPLEMENT THE ASSIGNED PROJECTS USING THE OPENCV SOFTWARE LIBRARIES.

IN ORDER TO PARTICIPATE TO THE FINAL ASSESSMENT AND TO GAIN THE CREDITS CORRESPONDING TO THE COURSE, THE STUDENT MUST HAVE ATTENDED AT LEAST 70% OF THE HOURS OF ASSISTED TEACHING ACTIVITIES.
Verification of learning
THE EXAM AIMS AT EVALUATING, AS A WHOLE: THE KNOWLEDGE AND UNDERSTANDING OF THE CONCEPTS PRESENTED IN THE COURSE, THE ABILITY TO APPLY THAT KNOWLEDGE TO SOLVE PROGRAMMING PROBLEMS REQUIRING THE USE OF ARTIFICIAL VISION TECHNIQUES; INDEPENDENCE OF JUDGMENT, COMMUNICATION SKILLS AND THE ABILITY TO LEARN.

THE EXAM INCLUDES TWO STEPS: THE FIRST ONE CONSISTS IN AN ORAL EXAMINATIONS AND IN THE DISCUSSION OF MID TERM PROJECTS REALIZED DURING THE COURSES. THE SECOND STEP CONSISTS IS BASED ON THE REALIZATION OF A FINAL TERM PROJECT: THE STUDENTS, PARTITIONED INTO TEAMS, ARE REQUIRED TO REALIZE A SYSTEM, FINALIZED TO A COMPETITION AMONG THE TEAMS, DESIGNING AND METHODOLOGICAL CONTRIBUTIONS OF THE STUDENTS, TOGETHER WITH THE SCORE ACHIEVED DURING THE COMPETITION, ARE CONSIDERED FOR THE EVALUATION.
THE AIM IS TO ASSESS THE ACQUIRED KNOWLEDGE AND ABILITY TO UNDERSTANDING, THE ABILITY TO LEARN, THE ABILITY TO APPLY KNOWLEDGE, THE INDEPENDENCE OF JUDGMENT, THE ABILITY TO WORK IN A TEAM.

IN THE FINAL EVALUATION, EXPRESSED IN THIRTIETHS, THE EVALUATION OF THE INTERVIEW AND OF THE MID TERM PROJECTS WORK WILL ACCOUNT FOR 40% WHILE THE FINAL TERM PROJECT WILL ACCOUNT FOR 60%. THE CUM LAUDE MAY BE GIVEN TO STUDENTS WHO DEMONSTRATE THAT THEY CAN APPLY THE KNOWLEDGE AUTONOMOUSLY EVEN IN CONTEXTS OTHER THAN THOSE PROPOSED IN THE COURSE.
Texts
LECTURE NOTES.

SZELISKI. “COMPUTER VISION: ALGORITHMS AND APPLICATIONS”, SPRINGER
M. SONKA, V. HLAVAC, R. BOYLE: "IMAGE PROCESSING, ANALYSIS AND MACHINE VISION", CHAPMAN & HALL.

THE TEACHING MATERIAL IS AVAILABLE ON THE UNIVERSITY E-LEARNING PLATFORM (HTTP://ELEARNING.UNISA.IT) ACCESSIBLE TO STUDENTS USING THEIR OWN UNIVERSITY CREDENTIALS.
More Information
THE COURSE IS HELD IN ENGLISH.
Lessons Timetable

  BETA VERSION Data source ESSE3