Computational imaging systems have variety of applications include consumer cameras, cell phone cameras, vehicle camera systems, surveillance, medical imaging, remote sensing, and human computer interaction. Topics of computational imaging have a wide range of technologies in computer vision and image processing. Recently, the network-based image processing become hot topic. This course focuses on the network-based image processing. In this course, students develop and train the network by themselves.
Students are expected to
(i) gain an ability to build and learn deep neural networks,
(ii) gain an ability to use numerical computing environments using MATLAB to solve engineering problems,
(iii) gain practical skill to apply the deep learning techniques such as momentum, data arugumentation and filter setting, after taking this course.
Computational imaging, Image processing, Convolutional neural network (CNN), Deep learning, matlab
This class is a kind of active learning. Instructor will give some information, but students are required to develop their own matlab code.
|第1回||Introduction of this course and of grader system.||Logical operation by neural network|
|第2回||Two-layer logical network & simple image processing||Two-layer logical network & simple image processing|
|第3回||Introduction of train1000 project: train the network with 1000 samples.||Train 1000 project|
|第4回||Introduction of BlockScramble challenge. Type-I: Supervised learning Type-II: Unspervised learning||Develop and train own network.|
|第5回||Key techniques of CNN||Develop and train own network.|
|第6回||Evaluation data submission||Evaluation data submission|
Presentation, and report.
Students will code by themselves.