As part of UASPIRE, you'll be developing an image processing system! To help you do that, we're going to be covering everything from basic image transformations to convolutional neural networks to pipelining.
This quarter, we'll be working on identifying UCLA building signs! The signs will be taken at weird angles, and we'll be giving you images with metadata related to the angles we take them at, e.g.
- Angle relative to the horizon
- Distance + Orientation relative to the Sign
Your job is to localize, classify, and rectify these images automatically.
Training Project Leads
- Ben Limpanukorn
- Andy Peng
Feel free to shoot either of us a message on Slack if you ever have any questions!
- Cost: Nothing! (Except effort, probably.)
- Groups of 2-3
Here are topics that will likely end up being covered in UAS-wide workshops:
Here are topics that we'll be covering but will be covered in more detail by other groups within UAS or other student organizations at UCLA:
- If you're interested in hardware, check out the Airframe/Ground/Controls UASPIRE workshops.
- If you're interested in web/app development, either frontend or backend, chect out ACM Hack's workshops.
- If you're interested in machine learning and AI, check out ACM AI's workshops. Ben (one of the subteam leads) is also involved with organizing this!
Here's a rough timeline for all of Fall Quarter, and a link to some slides with pictures if you'd prefer that. We'll be adding links to wiki pages/workshop slides as they happen.
Week 1 - 2: Introduction
- General kickoff meeting - Thurs Week 1
- Overview Meeting - Tues Week 2
Week 3: Basic Image Representation and Manipulation
- Basics, tools, PIL, OpenCV
- Image Operations and Filtering
Week 4: Image Processing and Basic Computer Vision
- Off-axis rectification
- Edge Detection
Week 5: Training Data Augmentation and Generation
- Target Generation
- Keras Data Augmentation
Week 6: Advanced Computer Vision
- Feature Detection
Week 7: Machine Learning
Week 8: Pipelining
Week 9: Competition