Abstract:

Since Google autonomous vehicles or self-driving cars became available in 2013, Computer Vision has become more popular. However, teaching Computer Vision as a second specialization, or a post-doctoral area can be very daunting in a small university. A small university is used here to refer to a university where teaching load is generally high (3-4 courses per semester) without offering a Ph.D. program such that no Ph.D. students can assist in research activities proactivity.

During the process, the experience of starting from “ground zero” literature study until completing the teaching in three semesters can be considered an incremental approach from the viewpoints of course design and student learning. From the viewpoint of course design, the course of Computer Vision is divided into a low-level Visual Computing course focusing on digital image processing; and an intermediate-level computer vision course that is still under development. From the pedagogical point of view, the teaching materials and the depth of coverage are incrementally enhanced.

From the view point of curriculum design, the subjects of computer vision can be covered seamlessly at two levels: low level and intermediate level. Low-level computer vision emphasizes topics about the fundamentals about image capturing, image processing, and some image analysis. The intermediate-level computer vision involves in image analysis. The incremental approach can be applied to cover the Visual Computing first with extended time to investigate the subjects at the intermediate level.

From the view point of student learning, the incremental approach also helps students to become more motivated. Learning Computer Vision requires some preparatory mathematics.
It is generally agreed that learning Computer Vision would be more effective if students complete the visual computing/digital image processing as a prerequisite course. For the low-level computer vision, one must have background on linear algebra, trigonometry, geometry, calculus, and digital signal processing. For the intermediate-level computer vision, students will learn machine learning methodologies with supervised & unsupervised learning.

As of today, the Visual Computer course has been developed and taught three times using the freeware OpenCV. The journey of curriculum design & professional development process took one year of preparation, three semesters of teaching just to become initially mature, and is still looking ahead for improvement. In this paper, the process and the experiences of teaching Visual Computing as the low-level Computer Vision course are reported. The timeline including the one-year preparation, detailed activities from the preparation stage, and the deployment stage are included. The student evaluation result & feedback collected after teaching the course the second time in Fall 2019 are discussed. The experiences learned and future direction are also reported.

Keywords:

Curriculum design, Computer Vision, Visual Computing, OpenCV.