Preliminaries

This week, I focused on refining and enhancing the deep learning exercises Human Detection and Digit Classification to improve their clarity, usability, and visual presentation. I updated the documentation for both exercises by adding relevant example code snippets, refreshing the instruction images, and creating step-by-step video tutorials to guide users through the process of running and interacting with the exercises.

Based on feedback provided by the reviewer, I also updated the hover effect of the upload button in both exercises to improve the UI/UX. To ensure proper tracking and collaboration, I created GitHub issues for these updates and submitted pull requests with the corresponding changes to the gh-pages branch.

In parallel, I followed the recommended "Deep Learning with PyTorch: A 60 Minute Blitz" tutorial by mentors. This hands-on study helped me strengthen my understanding of PyTorch fundamentals, which will be useful for future improvements and model development within the exercises.

Objectives

  • Implement an abstract model path feature to support dynamic ONNX model paths in exercises.
  • I updated and improved the web documentation for the Human Detection and Digit Classification exercises, enhancing clarity, structure, and user guidance.
  • Create GitHub issues to track documentation-related updates.
  • Submit pull requests to publish updated documentation.
  • Create a tutorial-style YouTube video and instruction images to showcase how users can interact with the exercises.
  • Improved the front-end interface of the exercises based on the review feedback
  • study and hands-on practice Deep Learning with PyTorch

Execution

Abstract Model Path

Implement an abstract model path feature to support dynamic ONNX model paths in exercises. Implement an abstract model path feature to support dynamic ONNX model paths in exercises. So the user can import the model path rather than the hard-coded path. If we change the model storage path internally, then we don't need to update all the documentation, and the user can use the path seamlessly.

Example Code:
                                    from model import model_path

Enhancing and Enrich document

I updated the documentation for both deep learning exercises. This includes adding example code snippets, updating the instruction images, and creating a tutorial video for each exercise to help users better understand and use the exercises. Additionally, I clarified the step-by-step workflow for running the exercises, highlighted common issues users might encounter, and provided troubleshooting tips. The documentation now also explains the expected input/output formats, and links to relevant resources for further learning. These improvements aim to make the exercises more accessible and user-friendly for both beginners and advanced users.

week-2-img-1 Basic instruction of digit classification
week-2-img-1 Basic instruction of human detection

GitHub Issues and Pull Requests

I created GitHub issues to track the updates made to the documentation of the Human Detection and Digit Classification exercises. These issues outline the specific changes made, such as adding example code snippets, updating instruction images, and creating tutorial videos. I also submitted pull requests with the corresponding changes to the gh-pages branch. These pull requests include detailed descriptions of the changes made, along with links to the updated documentation. The issues and pull requests serve as a record of the updates made to the documentation and provide a way for others to review and contribute to the documentation.

GitHub Issues:
GitHub PRs:

Tutorial Videos

I created a tutorial-style YouTube video for each exercise to showcase how users can interact with the exercises. These videos provide a step-by-step walkthrough of the exercises, demonstrating how to run the exercises, input data, and interpret the results. The videos are designed to be user-friendly and accessible, making it easier for users to understand and use the exercises effectively.

Human Detection Exercise Video
Digit Classification Exercise Video

Improve exercises UI

I updated the hover effect of the upload button in both the Human Detection and Digit Classification exercises, incorporating the feedback provided by reviewer.

Deep learning study

I have done the study and hands-on practice "Deep Learning with PyTorch: A 60 Minute Blitz," recommended by mentors.

References

[1] Deep Learning with PyTorch: A 60 Minute Blitz

[2] Deep learning-based Human Detection Exercise

[3] Deep learning-based Digit Classification Exercise