I’m eager to delve deeper into the fascinating worlds of machine learning and computer vision, building upon the solid foundation laid by the Udacity Self-Driving Car Nanodegree.
While the Nanodegree provided a whirlwind tour of these fields, I confess that I often felt like a kid in a candy store, overwhelmed by the sheer amount of information. To rectify this, I plan to revisit key concepts and tools, turning my surface-level understanding into a more profound knowledge.
My immediate goal is to bring real-time image recognition to life using YOLO or SSD. And, in a bold move, I aspire to hone my technical writing skills by crafting conference paper-worthy documents.
To accomplish these lofty ambitions, I’ve identified the following areas for improvement:
| Subject |
Things to learn |
Small goal |
| Technical writing |
Revisit how to write a technical paper |
- Complete a Coursera course |
| Unix/Git |
Understand commonly used Unix tool/command |
- Take a course to have general overview of Linux environment
- Install Ubuntu on my laptop for deep learning and computer vision |
| Image annotation |
utilize an existing tool to annotate the image |
- Practice to annotate the traffic signal images to prepare YOLO/SSD works |
| Data importing/exporting |
OPENCV
Python
Tensorflow |
- Take OpenCV course
- Take Python data science specialization course
- Take Coursera self driving car specialization course
- Take Tensorflow course
- Determine the best way to import and process traffic signal images and visualize the data importing process nicely |
| Data representation |
matplotlib |
- Do benchmark how other cohorts represented the image processing result using matplotlib
- Do go over codes and implemented in my own way for the later usage
- Review and implement in the same way how some of deep learning papers represent the data images |
| Vision sensing/image processing |
- OpenCV
- Python |
- Create a small program to recognize hand-writting image
- Make sure to visualize the steps of image processing |
| Deep learning |
OpenCV |
- Complete the OpenCV course
- Go back to the algorithm(undone from the last capstone) to detect traffic light from Bosch traffic signal data set |
| Deep learning |
Classification, not limited to CNN,YOLO,SSD based on Keras and Tensorflow |
- Create an algorithm to detect my daughter’s face in real time |