Introduction to Computer Vision and Its Applications
A.Y. 2025/2026
Learning objectives
This course introduces participants to the foundational concepts and methodologies of computer vision, with a particular focus on applications within the humanities studies. Through theoretical insights and practical tools, it aims to:
Present key artificial intelligence techniques, especially supervised deep learning models.
Explore computer vision tasks such as image classification, semantic segmentation, and text recognition.
Demonstrate how to prepare and process visual datasets for humanities research.
Foster an understanding of tools like Roboflow, Google Teachable Machine, and Transkribus for real-world applications.
Present key artificial intelligence techniques, especially supervised deep learning models.
Explore computer vision tasks such as image classification, semantic segmentation, and text recognition.
Demonstrate how to prepare and process visual datasets for humanities research.
Foster an understanding of tools like Roboflow, Google Teachable Machine, and Transkribus for real-world applications.
Expected learning outcomes
By the end of this course, students will be able to:
Understand Core Concepts: Explain the distinctions between AI, machine learning, and deep learning, and describe their relevance to computer vision.
Apply CV Techniques: Identify and implement basic computer vision tasks, such as image classification, using appropriate tools.
Develop Annotated Datasets: Create and manage labeled image datasets suitable for training classification models using platforms like Roboflow.
Experiment with AI Tools: Use Google Teachable Machine for training simple classifiers and interpret the results.
Process Historical Manuscripts: Apply convolutional neural networks for handwritten text recognition through tools such as Transkribus.
Critically Engage with Humanities Data: Reflect on the methodological challenges and opportunities of applying CV in digital humanities contexts.
Students will also have access to all materials via the myAriel platform and can consult the instructor during office hours or by email.
Understand Core Concepts: Explain the distinctions between AI, machine learning, and deep learning, and describe their relevance to computer vision.
Apply CV Techniques: Identify and implement basic computer vision tasks, such as image classification, using appropriate tools.
Develop Annotated Datasets: Create and manage labeled image datasets suitable for training classification models using platforms like Roboflow.
Experiment with AI Tools: Use Google Teachable Machine for training simple classifiers and interpret the results.
Process Historical Manuscripts: Apply convolutional neural networks for handwritten text recognition through tools such as Transkribus.
Critically Engage with Humanities Data: Reflect on the methodological challenges and opportunities of applying CV in digital humanities contexts.
Students will also have access to all materials via the myAriel platform and can consult the instructor during office hours or by email.
Lesson period: Second semester
Assessment methods: Esame
Assessment result: voto verbalizzato in trentesimi
Single course
This course can be attended as a single course.
Course syllabus and organization
Single session
Responsible
Lesson period
Second semester
Course syllabus
This 40-hour course introduces social science students to computational and visual approaches in digital humanities. Through hands-on practice, exploration of AI concepts, and applied case studies, students will gain both conceptual and technical foundations to critically explore how computational methods can be used for text and image-based cultural analysis.
The schedule is divided into four parts, with a high intensity on Part B.
PART A: Python for Text Analysis
Goal: Achieve a "simple text analysis" capability using Python.
· Introduction: Course overview, Colab basics, first Python scripts.
· Python Essentials: Variables, data structures (lists, dictionaries), control flow (loops, conditionals), and defining functions.
· Data Handling: Working with basic text files; introduction to core libraries (e.g., NumPy, Matplotlib for basic visualization).
· Text Processing: Tokenization, filtering (stop words), and frequency counts.
· Mini-Analysis: Conducting and reflecting on a basic analysis of a corpus (e.g., speeches, policy briefs, manifestos).
PART B: AI and Computer Vision Foundations with Image Classification
Goal: Conceptual and practical foundations of AI and Machine Learning.
· AI Context: Introduction to AI and Machine Learning, Applications in cultural context
· Types of Learning: Supervised, unsupervised, and reinforcement learning.
· Data for Vision: Image data (pixels, features), building and annotating image datasets.
· NN Basics: From perceptron to Artificial Neural Networks. Understanding basic learning: forward/backward propagation, gradient descent.
· Deep Learning Tools: Introduction to Convolutional Neural Networks (CNNs) and transfer learning concept.
· Hands-on Experience: Practical demonstration and use of a no-code AI tool for image classification (e.g., Teachable Machine). Visual inspection and interpretation of results.
PART C: Manuscript Transcription and HTR Tools
Goal: Hands-on experience with advanced text recognition workflows.
· Historical Data: Preprocessing and layout segmentation of historical manuscripts.
· OCR vs. HTR: Conceptual understanding and challenges in recognizing handwritten text.
· HTR Workflow: Introduction to a dedicated HTR platform (e.g., Transkribus): data uploading, model selection, training basics (demonstration).
· Application & Assessment: Demonstration of transcription output, quality assessment, and error analysis.
· In-Class Practice: Guided practical session on model training concepts using a sample dataset.
PART D: Student Presentations and Critical Reflection
Goal: Consolidate learning through application, critical presentation, and discussion.
· (Focus: Application): Presentation/Submission of Teachable Machine Application Exercise (Group Work, 10% of final grade).
· (Focus: Critical Analysis): Presentations of AI-based Academic Articles in Cultural domain (Individual Work, 20%of final grade).
· Peer discussion on opportunities and limitations of AI in cultural research. Course wrap-up and reflection.
The schedule is divided into four parts, with a high intensity on Part B.
PART A: Python for Text Analysis
Goal: Achieve a "simple text analysis" capability using Python.
· Introduction: Course overview, Colab basics, first Python scripts.
· Python Essentials: Variables, data structures (lists, dictionaries), control flow (loops, conditionals), and defining functions.
· Data Handling: Working with basic text files; introduction to core libraries (e.g., NumPy, Matplotlib for basic visualization).
· Text Processing: Tokenization, filtering (stop words), and frequency counts.
· Mini-Analysis: Conducting and reflecting on a basic analysis of a corpus (e.g., speeches, policy briefs, manifestos).
PART B: AI and Computer Vision Foundations with Image Classification
Goal: Conceptual and practical foundations of AI and Machine Learning.
· AI Context: Introduction to AI and Machine Learning, Applications in cultural context
· Types of Learning: Supervised, unsupervised, and reinforcement learning.
· Data for Vision: Image data (pixels, features), building and annotating image datasets.
· NN Basics: From perceptron to Artificial Neural Networks. Understanding basic learning: forward/backward propagation, gradient descent.
· Deep Learning Tools: Introduction to Convolutional Neural Networks (CNNs) and transfer learning concept.
· Hands-on Experience: Practical demonstration and use of a no-code AI tool for image classification (e.g., Teachable Machine). Visual inspection and interpretation of results.
PART C: Manuscript Transcription and HTR Tools
Goal: Hands-on experience with advanced text recognition workflows.
· Historical Data: Preprocessing and layout segmentation of historical manuscripts.
· OCR vs. HTR: Conceptual understanding and challenges in recognizing handwritten text.
· HTR Workflow: Introduction to a dedicated HTR platform (e.g., Transkribus): data uploading, model selection, training basics (demonstration).
· Application & Assessment: Demonstration of transcription output, quality assessment, and error analysis.
· In-Class Practice: Guided practical session on model training concepts using a sample dataset.
PART D: Student Presentations and Critical Reflection
Goal: Consolidate learning through application, critical presentation, and discussion.
· (Focus: Application): Presentation/Submission of Teachable Machine Application Exercise (Group Work, 10% of final grade).
· (Focus: Critical Analysis): Presentations of AI-based Academic Articles in Cultural domain (Individual Work, 20%of final grade).
· Peer discussion on opportunities and limitations of AI in cultural research. Course wrap-up and reflection.
Prerequisites for admission
No prior background in computer science is required; familiarity with cultural and historical sources (texts, images, artifacts) is expected, since these will inspire the applied projects. Basic mathematical literacy, ie., percentages, averages, reading simple charts, a touch of algebra, is helpful. A general awareness of AI from news or cultural debates is desirable (no technical knowledge assumed). While not mandatory, prior programming, statistics, or everyday digital tools (e.g., Excel, image editing) are welcome. Students should be willing to work in pairs, run guided Colab notebooks, and learn to interpret simple evaluation metrics. All essential Python will be taught from scratch.
Teaching methods
The course combines instructor-led sessions, where key concepts of artificial intelligence and neural networks are introduced, with guided practice in Python and classification tasks. Lectures and demonstrations will provide the necessary theoretical grounding, while Colab notebooks and case studies will allow students to apply these ideas step by step. As the course progresses, students move from structured exercises to collaborative project work, applying computational methods to cultural and historical materials. Active participation in discussions, pair work, and project development is strongly encouraged, since the course relies on continuous interaction between explanation and practice. Attendance, although not mandatory, is highly recommended to keep pace with both the conceptual and applied components.
Teaching Resources
- Slides, handouts, scholarly articles provided by the instructor.
- Brian Kokensparger, 2018, Guide to Programming for the Digital Humanities: Lessons for Introductory Python
- Think Python: How to Think Like a Computer Scientist by Allen B. Downey (Second Edition, 2015), covering the key topics discussed in lectures. https://open.umn.edu/opentextbooks/textbooks/43, https://greenteapress.com/wp/think-python-2e/
- A Beginner's Guide to Python 3 Programming by J. Hunt (https://minerva.unimi.it/permalink/39UMI_INST/i9q3jt/alma991017213265306031)
- NumPy (https://numpy.org/) and pandas (https://pandas.pydata.org/) have excellent documentation online.
- Ian Goodfellow, Yoshua Bengio, Aaron Courville, Deep Learning (Adaptive Computation and Machine Learning series), online: https://www.deeplearningbook.org/
- David G. Stork, 2023, Pixels & Paintings: Foundations of Computer-assisted Connoisseurship
- Brian Kokensparger, 2018, Guide to Programming for the Digital Humanities: Lessons for Introductory Python
- Think Python: How to Think Like a Computer Scientist by Allen B. Downey (Second Edition, 2015), covering the key topics discussed in lectures. https://open.umn.edu/opentextbooks/textbooks/43, https://greenteapress.com/wp/think-python-2e/
- A Beginner's Guide to Python 3 Programming by J. Hunt (https://minerva.unimi.it/permalink/39UMI_INST/i9q3jt/alma991017213265306031)
- NumPy (https://numpy.org/) and pandas (https://pandas.pydata.org/) have excellent documentation online.
- Ian Goodfellow, Yoshua Bengio, Aaron Courville, Deep Learning (Adaptive Computation and Machine Learning series), online: https://www.deeplearningbook.org/
- David G. Stork, 2023, Pixels & Paintings: Foundations of Computer-assisted Connoisseurship
Assessment methods and Criteria
For attending students, the overall assessment of the theoretical and applied skills acquired will be based on a final oral examination focusing on learning materials provided (50%), on the preparation of individual and collective work (30%), and on the active participation during classes (20%). Self-assessment tools may be provided to attending students during the course. For non-attending students, the knowledge and competences developed will be assessed by a final oral examination focusing on learning materials provided (50%), and the preparation of individual or collective work (50%).
The final oral exam consists of an oral discussion, designed to assess the students' understanding of the main concepts introduced during the course, their ability to explain with clarity and to link the different topics and issues addressed, their capacity to analyse case studies with appropriate theoretical tools and critical awareness.
The final oral exam consists of an oral discussion, designed to assess the students' understanding of the main concepts introduced during the course, their ability to explain with clarity and to link the different topics and issues addressed, their capacity to analyse case studies with appropriate theoretical tools and critical awareness.
Professor(s)