Overview

Course Description

• This is a Special Topics course focusing on the foundations of computer vision and machine learning/ deep learning methods for computer vision tasks. The students are expected to have obtained a solid background in machine learning, linear algebra, and coding skills.
• This course is a combination of lectures and student presentations. We plan to read and discuss recent research papers on the intersection between computer vision and machine learning. We plan to invite internal (within the department) and external speakers to give guest lectures to give introductory and state-of-the-art developments in the area of computer vision and machine learning.
• This course satisfies the ECE graduate program requirement.
• Piazza: https://piazza.com/ubc.ca/winterterm22022/eece570

Contacts

• Instructor: Dr. Xiaoxiao Li (xiaoxiao.li@ece.ubc.ca)
• TA: Chun-Yin Huang (chunyinhuang17@gmail.com)
• TA: Sana Ayromlou (ayromlous@gmail.com)
• Email: include ‘[EECE 570]’ in the subject.

Time and Location

• Location: Tue/Thu ∥ 11:00 am – 12:30 pm ∥ SPPH B151
• Zoom participation ID: https://ubc.zoom.us/j/65807504716?pwd=RnV0Z2dHR1NCMkcwcnVYVmluZ2RRdz09
• Zoom meeting ID and Password: 658 0750 4716 ; 945382
• TA Office Hours: TBA
• Instructor Office Hours: Thursday afternoon (by appointment only)

Reading and Presentation

• During each class meeting, we will have student presentations and discussions of selected papers. The tentative schedule below lists tentative topics for each class and suggested papers.

• Students taking the course are expected to read all selected papers. The presenter needs to write a review for the paper to be presented. Each review should be submitted one day before its presentation and discussion. Please refer to lecture 1’s course notes on how to write a review. The review needs to cover 1) summary; 2) pros; 3) cons; and 4) future direction.

• There are 3 parts for each presentation:
  • Problem statement, a brief overview of the state-of-the-art
  • Key idea of the paper
  • Strengths, weaknesses and future directions

• The total time allocated for each paper is 40 minutes with 25 mins presentation and 15 mins discussion. We suggest you prepare in advance and do not exceed 40 minutes to leave 40 minutes for another presentation.

• We welcome students to suggest their interested papers. If students are interested in substitute papers, after signing up for desired time slots, the student should discuss with the instructor at least one week in advance to have time to make an announcement.

Grading

• Paper Readings and Presentations: 30%
  • Read the selected paper(s) and submit review
  • Present the selected paper
  • Lead discussion in the classroom
• Course participants: 10%
  • You must show up on your presentation date. No show will lead to failing the course.
  • Grade for your classmate’s presentation. If you cannot attend the presentation (two chances over the term), please let TA and instructor know. Otherwise, -1 point each time.
• Project proposal (2 pages): 20%
• Final project (at least 4 pages): 40%
  • A computer vision project that is related to the course topic. You have the flexibility to choose the topics. More details will be released later. No Teamwork allowed.
  • Passing the course does on conditional on if you pass the final project.
• Late submission will result in *0.8 decay per day. Extension is only accepted via applying for Academic Concession.

Computational Resources

• GPU computing is required for this class. I strongly recommend to Google Colab or use your own/lab’s GPU since that is the most convenient way of writing and testing code with GUI. Click here to try out the Colab tutorial. You can also consider using other computational resources offered by UBC.

Optional Textbooks

• Friedman, Jerome, Trevor Hastie, and Robert Tibshirani. The elements of statistical learning. Vol. 1. No. 10. New York: Springer series in statistics, 2001.
• Richard Szeliski. Computer Vision: Algorithms and Applications , 2022.
• Goodfellow, Ian, Yoshua Bengio, Aaron Courville, and Yoshua Bengio. Deep learning. Vol. 1, no. 2. Cambridge: MIT press, 2016.
• Torfi, Amirsina. Deep Learning Roadmap

---

Schedule

Our schedule will be updated during the semesber. Please frequently check the schedule here.

Dates	Presenters	Topics	Suggested papers	Submissions
1/10	Dr. Xiaoxiao Li	Couse Introduction Recording Password: %Ug4qW1+		Signup Piazza
1/12	Dr. Xiaoxiao Li	Introduction to computer vision slides
1/17	Dr. Xiaoxiao Li	Introduction to deep learning slides
1/19	Dr. Xiaoxiao Li	Edge Detector slides
1/24	Luke Ruichen	Presentation – deep learning-based sketching	Chan, C., Durand, F. and Isola, P., 2022. Learning to generate line drawings that convey geometry and semantics. CVPR 2022 paper He, J., Zhang, S., Yang, M., Shan, Y. and Huang, T., 2019. Bi-directional cascade network for perceptual edge detection. CVPR 2019 paper	Submit review Submit peer-grading link
1/26	Dr. Xiaoxiao Li	Fileters and Multi-scale Representation		Signup presentation
1/31	Dr. Xiaoxiao Li	Image Classification I and Object Detection
2/2	Danni, Hongrong peer-review Yingrui, Yiming peer-review	Presentation – Multi-scale image classification	Huang, G., Chen, D., Li, T., Wu, F., Van Der Maaten, L. and Weinberger, K.Q., 2017. Multi-scale dense networks for resource efficient image classification. ICLR 2018 paper Pang, Y., Wang, T., Anwer, R.M., Khan, F.S. and Shao, L., 2019. Efficient featurized image pyramid network for single shot detector. CVPR 2019 paper	Submit review Submit peer-grading
2/7	Dr. Xiaoxiao Li	Image Classification II – Practical Callenges
2/9	Larry, Weige peer-review Mingyuan peer-review	Presentation – Advanced image classification	Li, J., Socher, R. and Hoi, S.C., 2020. Dividemix: Learning with noisy labels as semi-supervised learning. ICLR 2020 paper Kang, J., Lee, S., Kim, N. and Kwak, S., 2022. Style Neophile: Constantly Seeking Novel Styles for Domain Generalization. CVPR 2022 paper
2/14	Dr. Xiaoxiao Li	Image Classification III – Trustworthiness
2/16	Mohammad peer-review Mingrui, Victor peer-review	Presentation – Advanced image classification	Doan, K.D., Lao, Y. and Li, P., 2022. Marksman Backdoor: Backdoor Attacks with Arbitrary Target Class. NeurIPS 2022 paper Rigotti, M., Miksovic, C., Giurgiu, I., Gschwind, T. and Scotton, P., 2021, September. Attention-based Interpretability with Concept Transformers. ICLR 2022 paper
3/3	(updated deadline)			Submit project proposal
2/28	Dr. Xiaoxiao Li	Vision-Transformer
3/2	Chunyu, Ali peer-review Ningyuan, Jiahe peer-review	Presentation – Vision Transformer	Ding, M., Xiao, B., Codella, N., Luo, P., Wang, J. and Yuan, L., 2022. DaViT: Dual Attention Vision Transformers. ECCV 2022 paper Sun, T., Lu, C., Zhang, T. and Ling, H., 2022. Safe Self-Refinement for Transformer-based Domain Adaptation. CVPR 2022 paper	Submit review Submit peer-grading
3/7	TEA grad student	Trustworthy AI
3/9	Dr. Fatemeh Dezaki (Amazon)	Reasech topic in computer vision
3/14	Dr. Xiaoxiao Li Chenyu You (Yale)	Image Segmentation
3/16	Dr. Renjie Liao	Image Synthesis
3/21	Chen, Weiya peer-review Baichuan, Kevin peer-review	Presentation – Advanced image segmentation	Isensee, F., Jaeger, P.F., Kohl, S.A., Petersen, J. and Maier-Hein, K.H., 2021. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature methods paper Ru, L., Zhan, Y., Yu, B. and Du, B., 2022. Learning Affinity from Attention: End-to-End Weakly-Supervised Semantic Segmentation with Transformers. CVPR 2022 paper	Submit review Submit peer-grading
3/23	Beidi, Yilin peer-review Christina, Yu Gao peer-review	Presentation – Advanced image synthesis	Park, T., Efros, A.A., Zhang, R. and Zhu, J.Y., 2020, August. Contrastive learning for unpaired image-to-image translation. ECCV 2020 paper Chung, H., Sim, B., Ryu, D. and Ye, J.C., 2022. Improving Diffusion Models for Inverse Problems using Manifold Constraints. NeurIPS 2022 paper	Submit review Submit peer-grading
3/28	Dr. Xiaoxiao Li Bo Zhou (Yale)	Medical image registration and reconstruction
3/30	Dr. Xiaoxiao Li	Video Understanding
4/4	Dong, Li peer-review Jackson, Kevin peer-review	Presentation – Image registration	Chen, J., Frey, E.C., He, Y., Segars, W.P., Li, Y. and Du, Y., 2022. Balakrishnan, G., Zhao, A., Sabuncu, M.R., Guttag, J. and Dalca, A.V., 2019. VoxelMorph: a learning framework for deformable medical image registration. IEEE TMI paper Zhou, B., Schlemper, J., Dey, N., Salehi, S.S.M., Sheth, K., Liu, C., Duncan, J.S. and Sofka, M., 2022. Dual-domain self-supervised learning for accelerated non-Cartesian MRI reconstruction. Medical Image Analysis paper	Submit review Submit peer-grading
4/6	Ailar peer-review Parsa, Mobina peer-review	Presentation – Video Analysis	Isobe, T., Li, S., Jia, X., Yuan, S., Slabaugh, G., Xu, C., Li, Y.L., Wang, S. and Tian, Q., 2020. Video super-resolution with temporal group attention. CVPR 2022 paper Dave, I.R., Chen, C. and Shah, M., 2022. SPAct: Self-supervised Privacy Preservation for Action Recognition. CVPR 2022 paper	Submit review Submit peer-grading
4/11	Dr. Xiaoxiao Li	Multimodel Image-text Classification
4/21	All students	Final project		Submit final project

---

Commitments

• It is my ultimate goal for this course, and my teaching, to develop your academic skills, supporting your future study and career. To do so will require commitments from you and myself toward meeting this goal.

Active Participation

• I will be prepared to use class time to help you understand the course material. I will respectfully listen to, understand, and answer questions asked in class.
• You are expected to attend class and actively participate in discussions, answering questions, asking questions, presenting material, etc. Your participation will be respectful of your classmates, both of their opinions and of their current point in their educational journey, as we each approach the material with different backgrounds and contexts.

Academic Integrity

• This course follows the rules presented in Acdemic Integrity at UBC.

Learning Accomodation

• I will make this classroom an open and inclusive environment, accommodating many different learning styles and perspectives.
• Any student seeking accommodation in relation to a recognized disability should inform me at the beginning of the course.

Physical and Mental Health

• I am willing to work with you individually when life goes off the rails.
• Coursework and college in general can become stressful and overwhelming, and your wellness can be impacted when you least expect it. You should participate in self-care and preventative measures, and be willing to find support when you need it.