Course Information for Spring 2016
Lab Instructor Information
Prof. Bruce A. Maxwell
Office hours: Knock
Investigates designing computer programs that extract information from digital images. Major topics include image formation and acquisition, gray-scale and color image processing, image filters, feature detection, texture, object segmentation, classification, recognition, and motion estimation. Students are introduced to classic and contemporary vision techniques with examples for homework and programming assignments drawn from biological and medical imaging, robotics, augmented reality, and digital photography. Students will develop small and medium-scale vision systems to solve practical problems and possibly assist in active research projects at Colby.
- Students understand the fundamentals of image formation and image acquisition.
- Students understand and can implement image processing routines used in computer vision algorithms, such as filtering and morphological operations.
- Students can discuss and implement algorithms for feature detection, segmentation, classification, and tracking.
- Students work in a group to design and develop a medium-sized image analysis and computer vision application.
- Students present algorithms and results in an organized and competent manner, both written and orally.
There are no great computer vision textbooks. There are good computer vision textbooks that are somewhat old (Stockman and Shapiro, or Sonka and Hlavac). There is a reasonable computer vision text that is free in electronic form (Szeliski). There are a number of reference style texts, mostly covering the software OpenCV and its various language APIs. You can access a decent OpenCV reference in the Colby Library Safari Online service. I would recommend downloading the Szeliski Book and using the OpenCV reference to get both the theoretical and practical side of computer vision.
- Maxwell's Lecture Notes
- Final: Deza and Parikh, "Understanding Image Virality", CVPR 2015.
- Final: Nguyen, Yosinksi, and Clune, "Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images", CVPR 2015.
- OpenCV: computer vision SDK.
- CVOnline: computer vision resources.
- Computer Vision Foundation: computer vision papers and community news.
- Computer Vision Papers: more CV papers.