DeepLearningPetIdentification

This project focuses on comparing images of lost pets with those already stored in a database. It aims to help identify and match lost pets with photos in a pre-existing dataset.

Project Structure

• scraper.py: A script used to scrape pet images and data from the website www.psi-detective.cz.
• pets_db.json: A database containing scraped pet data. Each entry includes a pet ID, description, and links to the corresponding images.
• utility.py: A collection of useful functions that can be utilized across multiple notebooks within the project.
• Jupyter Notebooks: Several notebooks, each implementing a different approach to pet classification. These notebooks demonstrate various methods for training and testing models to identify pets.

Collaboration Workflow

This document outlines the workflow for contributing to the project. Please follow the guidelines to ensure smooth collaboration and version control.

Table of Contents

1. Branch Naming Convention
2. Development on Local Machine
3. Development on Google Colab
4. Main Branch Protection
5. Pull Requests
6. Working with Pets Database
7. Utility Functions

1. Branch Naming Convention

Before every modification you should create a new branch first. When creating a new branch, follow this naming format:

• Feature Branch: feat_<feature_name>_<short_description>
• Fix Branch: fix_<issue_name>_<short_description>

This naming convention helps keep branches organized and easy to understand.

2. Development on Local Machine

If you are working on your local machine:

1. Clone the repository:

   git clone https://github.com/your-username/your-forked-repository.git

2. Checkout to your newly created branch:

   git checkout feat_image_preprocessing_resize

3. Continue your development in a classic Git workflow:
   • Edit files.
   • Stage your changes:

     git add .

   • Commit your changes:

     git commit -m "Description of changes"

   • Push your changes to the remote branch:

     git push origin feat_image_preprocessing_resize

3. Development on Google Colab

If you are using Google Colab:

1. Create a new Colab notebook.
2. Save a copy of the notebook to your GitHub repository (select the branch you are working on).
   • In Colab, click on File → Save a copy in GitHub.
   • Choose the repository and the appropriate branch (e.g., feat_image_preprocessing_resize).
3. Work within the notebook and push your changes back to GitHub when finished.

4. Main Branch Protection

The main branch is protected and can only be updated via pull requests (PRs). This ensures that the main codebase remains stable and any changes are reviewed before merging.

5. Pull Requests

When you are ready to submit your work:

1. Push your changes to your remote branch (e.g., feat_image_preprocessing_resize).
2. Open a pull request (PR) from your branch to the main branch.
3. PR Approval: Every pull request must be approved by one additional team member. This ensures the code is reviewed for quality, accuracy, and completeness.
4. Descriptive Comments: Each PR should include a descriptive comment explaining what was changed, why, and how it works.

6. Working with Pets Database

In the GitHub repository, there is a pets_db.json file that contains scraped data on pets.

• Issue: The dog breed "křižovec" incorrectly includes some cat entries.
• Solution: If you are fine-tuning the model or using the database, you should probably exclude all lines with the breed "křižovec" to avoid this issue.

7. Utility Functions

The utility.py file contains useful functions that can be reused across different notebooks.

• If you create a new function that could be useful for other notebooks, add it to the utility.py file.
• Make sure to provide a clear description of the function in the docstring so others can understand its purpose and how to use it.

Important: Updating utility.py

To use the utility functions in your notebooks, copy the top cells from PetsDetection.ipynb. These cells download utility.py and import its functions for use:

# Download utility.py file
import requests

# Correct raw URL for the utility.py file
url = "https://raw.githubusercontent.com/avkaz/DeepLearningPetIdentification/main/utility.py"

# Fetch and save the file locally
response = requests.get(url)
with open("utility.py", "wb") as f:
    f.write(response.content)


# Import utility functions
from utility import resize_image, another_function
import utility
print("utility.py downloaded successfully.")