This README provides instructions for setting up the Python environment for this project using VSCode.

• VSCode installed on your system
• Python 3.9 installed
• uv package manager installed, visit this for more information: click here

1. Clone the repository

   git clone https://github.com/tcstrength/item-identification.git
   cd item-identification

2. Set up Python environment with uv

   Create and activate a virtual environment:

   uv venv

   On Windows, activate the environment:

   .venv\Scripts\activate

   On macOS/Linux, activate the environment:

   source .venv/bin/activate

3. Install dependencies

   uv pip install -r requirements.txt

4. Configure environment variables

   Create a .env file in the root directory of the project:

   touch .env

   Add the following environment variables to the .env file:

   LABEL_STUDIO_URL=<your-label-studio-url>
   LABEL_STUDIO_API_KEY=<your-api-key>
   LABEL_STUDIO_TEMP_DIR=local/temp
   LABEL_STUDIO_PROJECT_MAPPING={"train":1,"validation":7,"test":3}
   

5. Configure VSCode

   • Open the project in VSCode
   • Install recommended extensions (Python, Python Environment Manager)
   • Select the Python interpreter:
     • Press Ctrl+Shift+P (Windows/Linux) or Cmd+Shift+P (macOS)
     • Type "Python: Select Interpreter"
     • Choose the interpreter from your .venv directory

• Make sure to never commit your .env file to version control
• If you update dependencies, update requirements.txt with: uv pip freeze > requirements.txt

• Selective Search for Object Detection: https://www.geeksforgeeks.org/selective-search-for-object-detection-r-cnn/
• YOLOv12: https://docs.ultralytics.com/models/yolo12/#citations-and-acknowledgements
• Low Object Count: https://chatgpt.com/share/6804ca08-de5c-8001-9be8-402becd967b8
• Contour-based Approach: https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=16b4468ae7e0e0193aa8593f8b3d4475b80c87e5