Upon successful completion of this course, students will be able to:
1. Demonstrate a solid conceptual understanding of how visual information can be processed by humans and machines.
2. Identify the challenges involved in understanding, modeling, and simulating human vision mechanisms.
3. Compare and contrast the human vision system and computer vision systems designed to achieve comparable functionality.
4. Interpret, summarize and critique scientific papers in the field of vision science, analyzing the implications and applications of their research findings.
5. Contextualize the latest advances in computer vision, artificial intelligence, and deep learning and their impact on the advancement of vision science.
Course description
This course provides a systematic overview of the interdisciplinary science of vision, combining psychological, neurophysiological, and computational aspects of vision research. It covers meaningful open problems in human and computer vision and discusses selected approaches for solving such problems. It also presents correlations between vision science and selected aspects of cognitive neurosciences, artificial intelligence, and art.
Outline / Topics
1. Introduction to vision science: history, methods, theories, approaches, visual perception
2. The human visual system: the human eye, the retina, receptive fields, brain cells, the visual cortex, brain maps, physiological mechanisms, psychophysical channels
3. Psychophysical methods in vision science: thresholds, sensitivity, bias, just noticeable difference, signal detection theory, two-alternative forced choice (2AFC) methods, Weber's, Fechner's and Stevens's laws
4. Spatial vision: edge detection, depth perception, size and distance estimation, the figure/ground problem, lightness perception (brightness, contrast), texture analysis, shape from texture, stereoscopic information, aftereffects
5. Color vision: physics of light and color, theories of color vision, physiological mechanisms, color constancy, color blindness, color naming
6. Object detection, recognition, and categorization: object properties and parts, representation of shape and structure, perception of function, theories of object categorization
7. Organizing objects and scenes: scene analysis, scene classification, objects in context
8. Visual dynamics: image motion, object motion, self-motion and optical flow
9. Visual selection, attention, search, and saliency: eye movements, bottom-up versus top-down visual attention, feature integration theory, visual search, computational models of visual saliency, applications
10. Visual memory and imagery: iconic, short- and long-term memory, visual imagery
11. Visual awareness: philosophical foundations, neuropsychology of visual awareness, famous patients (and what can be learned from them), theories of consciousness
12. Art and the brain: using works of art to learn more about the neurophysiology and psychology of vision
13. Deep Learning: basics of neural networks, fundamentals of deep learning techniques and algorithms, deep learning frameworks
14. Recent advances in computer vision using deep learning techniques: the end of the AI winter, deep learning everywhere, solved problems, open problems, achievements, promising directions for future research.
Course Assignments and Grading Scheme
The final grade will be based on the following weighted distribution.
Paper Review 35%
Problem Set 55%
Participation 10%
The Paper Review will require students to read a scientific paper in the field of vision science (chosen by the instructor) and produce a short and insightful summary / critique of its contents.
The Problem Set is essentially a collection of questions related to the topics covered in the class, similar to a take-home exam.
Participation points will be assigned based on the amount and quality of the student's contributions to discussions, in class and online.
