Visual Recognition
Code: 108268Credits: 6
| Degree programme | Type | Course |
|---|---|---|
| Bachelor in Artificial Intelligence | OP | 3 |
Contact lecturer
- Name :
- Debora Gil Resina
- Email :
- debora.gil@uab.cat
Group languages
You can consult this information at the end of the document.
Prerequisites
Have taken the subjects of Fundamentals of Machine Learning, Fundamentals of Programming, Fundamentals of Computer Vision, Probability and Statistics, and Neural Networks and Deep Learning.
It is recommended that the student has knowledge and skills of:
- Programming in the Python programming language
- Signal, Image and Video Processing
- Statistical invoicing
- Computational Learning and Deep Learning
Objectives
Every decade or so, there's a technological tsunami that transforms multiple industries. Artificial Intelligence (AI) is this wave that is sweeping the technological world today. If you've ever wondered:
How do computers perform face detection in crowds? How do video calling apps blur the background or replace the background with other images? How do self-driving cars move safely in an urban environment? How do you follow the ball so accurately at televised sporting events such as tennis, football and how do you follow the ball so accurately at televised sporting events such as tennis, football and basketball? Can we know the most effective cancer treatment from multimodal patient data? Can we know a person's emotions with a video? How do machines learn?
If we have piqued your curiosity, this course is the one you need. In this course we will depend on topics in Computer Vision such as Object Tracking, Image Classification, Personalized Medicine, Vision Language models and many more.
Unlike other computer vision courses, this course approaches computer vision in a more practical, experiential, and intuitive way. Its main component is a set of projects that must be developed by the students divided into teams. All that is needed is a working knowledge of the Python programming language. We will use python that allows us to incorporate different computer vision libraries. It's used by thousands of companies, products, and devices and tested every day for scalability and performance. In addition, we will depend on designing and adapting specific networks and choosing the most appropriate processing method according to the requirements and constraints of each application. In short, Vision and Learning is an eminently practical and interdisciplinary subject that is located on the bridge between artificial intelligence and the real world and that aims to bridge this bridge in both directions.
Learning outcomes
- CM12 (Build solutions to complex artificial intelligence problems based on the selection of the most appropriate image processing and computational learning techniques applied to computer vision.) Build solutions to complex artificial intelligence problems based on the selection of the most appropriate image processing and computational learning techniques applied to computer vision.
- CM13 (Plan the development and deployment of solutions based on computer vision using the appropriate tools and platforms, taking into account sustainability criteria in the use of computational resources.) Plan the development and deployment of solutions based on computer vision using the appropriate tools and platforms, taking into account sustainability criteria in the use of computational resources.
- KM29 (Identify the mathematical and algorithmic underpinnings on which low-level image processing, optimisation, and computational learning techniques applied to computer vision are based.) Identify the mathematical and algorithmic underpinnings on which low-level image processing, optimisation, and computational learning techniques applied to computer vision are based.
- KM30 (Identify the most common deep learning architectures and models for solving computer vision problems) Identify the most common deep learning architectures and models for solving computer vision problems
- SM31 (Apply the right machine learning techniques to machine vision problems.) Apply the right machine learning techniques to machine vision problems.
- SM32 (Use existing image processing algorithms and techniques, as well as deep learning architectures and models for the design and implementation of computer vision systems for different types of visual recognition problems.) Use existing image processing algorithms and techniques, as well as deep learning architectures and models for the design and implementation of computer vision systems for different types of visual recognition problems.
- SM33 (Perform data preparation, training, and model validation and analysis on all types of machine vision problems.) Perform data preparation, training, and model validation and analysis on all types of machine vision problems.
Contents
- Introduction to Computer Vision Computational Learning
- Image Classification
- Multi-Instance Learning
- Mechanisms of care
- Image Generation
- Multimodal Learning
- Foundation Models
- Vision Language Models
- few-shot learning
Learning activities and methodology
| Title | Hours | ECTS | Learning outcomes |
|---|---|---|---|
| Project Solving | 37.5 | 1.5 | CM12, CM13, KM29, KM30, SM31, SM32, SM33 |
| Individual work | 95 | 3.8 | CM12, CM13, KM29, KM30, SM31, SM32, SM33 |
| Lectures | 12.5 | 0.5 | CM12, CM13, KM29, KM30 |
The course will follow a teaching methodology of learning called Project-Based Learning (PBL).
The PBL methodology aims to empower and motivate the student in their learning. Groups of between 2 and 4 students will be formed and will be entrusted with carrying out a set of projects (of medium size) throughout the semester. There will be weekly monitoring and both group and individual tutoring of the students
The projects are set by the teaching staff in such a way that they meet the following conditions: they are as real as possible; be treatable by means of elementary tools; not have a standard solution algorithm associated with it. On the other hand, it's essential to understand that it's not about finding an algorithm that works 100 x 100 of the times – there is often no such thing – but simply about "giving you a reasonable solution proposal".
The projects have to be developed by each team with the maximum possible autonomy. Each team will be assigned a tutor who will follow its evolution but in principle will refrain from imposing its ideas. On the other hand, the student must be clear that it is not a matter of looking for the 'solution' of the problem elsewhere, but of making an original contribution. This does not mean that you have to give up the information that may exist in the bibliography or on the Internet; but when it is used, the teacher must be kept informed and explained in the memory.
Assessment
Continuous assessment activities
| Title | Weight | Hours | ECTS | Learning outcomes |
|---|---|---|---|---|
| Individual Mark | 30 | 5 | 0.2 | CM12, CM13, KM29, KM30, SM31, SM32, SM33 |
| Group Mark | 70% | 0 | 0 | CM12, CM13, KM29, KM30, SM31, SM32, SM33 |
Evaluation Activities
The subject has 2 assessment activities:
- Group projects (70%)
- Individual written test (30%)
Project Evaluation.The subject has 3 projects of increasing difficulty. At the end of each project, students will submit a report of the work carried out that will be evaluated by the professors of the subject, whether or not they are the tutors. The following INSTRUMENTS and ACTIVITIES will be used for the evaluation:
- PROJECT REPORT The report must follow the model provided in the CV. The degree of technical difficulty, reproducible experimental design, analysis of metrics and interpretation of results will be assessed. The report must include a section specifying what the contribution of each member of the group is and what has been made of the AI. For each project, the use of specific cheeks may be requested that constitute basic knowledge and skills that the student must have acquired in the resolution of the proposed problem. The code must allow a use case to be executed and can be used to verify that students have correctly developed the project,
- APPLICATION: developed program.
- CLASS MONITORING: Evaluation based on the observations made by the tutors in the tutored sessions, where the attitude, initiative, participation, attendance and punctuality of the student in the group sessions will be taken into account.
Given that the projects are developed throughout the course, their evaluation is continuous, and their final result is not recoverable.
Individual Test. At the end of the course, an individual written test will be taken where the student must demonstrate that they have understood the contents and methodologies used in the projects carried out.
Qualifications
The grade of the subject is the weighted average between the grade of the projects and the grade of the individual test:
Final grade = 0.7 * Project grade + 0.3 * individual grade
The grade of the projects will come out of the average of the 3 projects carried out. You have to have a 5 in the Individual Grade to make the average. The subject is approved if the Final Grade >=5
To distinguish between 'failed' and 'not presented', a deadline is set for students to withdraw from the assessment, in which case they will appear as 'not presented'. To unsubscribe, it will be necessary to notify the teacher, in writing or by email, and obtain a receipt acknowledgement.
If it is proven that some of the content of the project has been plagiarized and/or elaborated by a third person other than the student and/or generated by AI, it will be automatically suspended.
Single Assessment: This subject does not contemplate the single assessment system.
Use of AI. AI tools can be used as tools to support learning (e.g. to improve writing, style, clarity, linguistic correctness/hearing to obtain assistance in technical aspects). In no case may they replace and/or saturate the student's learning activity, or their acquisition of the specific knowledge of the subject.
It is not acceptable to use artificial intelligence tools to generate work content that is subject to evaluation. Evaluable tasks/activities suspected of having been generated by an AI instead of by the student will be considered as a copy and will be evaluated with a 0.
Bibliography
To be given
Software
Python, pytorch
Course groups and languages
The information provided is provisional until November 30. After this date, you will be able to consult the language of each group through this link. To access the information, you will need to enter the course CODE
| Type of teaching | Group | Language | Semester | Shift |
|---|---|---|---|---|
| (TE) Theory | 1 | English | first semester | afternoon |
| (PAUL) Classroom practices | 1 | English | first semester | afternoon |
| (PLAB) Practical laboratories | 1 | English | first semester | afternoon |