| Instructor | Akshat Dave, Assistant Professor [email protected] | | --- | --- | | Class location | NCS 115, New Computer Science | | Class hours | 5:00 PM - 6:20 PM Mondays and Wednesdays | | Brightspace Link | https://mycourses.stonybrook.edu/d2l/home/2252610 | | Office hours | NCS 153, 4 PM - 5 PM, Fridays |
<aside>
</aside>
Computational photography, also referred to as computational imaging, is an emerging field that combines optics, graphics, vision, and machine learning to capture and reconstruct visual information far beyond the reach of traditional cameras. From instruments that image black holes to self-driving cars that see around corners, computational photography is essential to recent advances in science, medicine, and consumer technologies.
In this graduate-level special topics course, we will explore the core principles and recent research in computational photography. Topics will include single-photon imaging, polarization-based vision, event cameras, and imaging through scattering. The course will cover both hardware innovations, such as single-photon sensors, and computational techniques, such as physics-aware neural representations, that enable imaging and perception in extreme scenarios.
Class format will be discussion based. Students will read, present, and discuss research papers, and complete a semester-long, research-oriented project.
By the end of the course, students will: • Develop a strong understanding of core concepts in computational photography – covering both camera hardware and computational components. • Critically read and present research papers that comprise (1) core advances in optics, sensors and illumination and (2) applications in robotics, biomedical imaging and astronomy. • Identify open research problems and explore them through a semester-long project.
A strong undergraduate or graduate background in at least one of the following is recommended: computer vision (e.g., CSE 327 or CSE 527), computer graphics (CSE 334 or CSE 528), image/signal processing (ESE 305 or ESE 342), or inverse problems / numerical methods (AMS 326/570 or ESE 589). You should also be comfortable with linear algebra and basic probability/statistics (MAT 211; AMS 210/310 or equivalents).
Following books are not required for the course, but are highly recommended for deeper exploration.
After Modules 1 and 2, each class will be discussion-based: each student will give a 45-minute presentation and lead a 30-minute class discussion.