tehwenyi / sahi_general

SAHI General

Requirements

• SAHI : 0.10.5
  pip install sahi
• CV Model : requires the below functions

  • classname_to_idx : int
    

    Parameters
    ----------
    classname : str
      class name of object
        
    Returns
    -------
    int
      index of the classname 
    

  • get_detections_dict :
    

    Parameters
    ----------
    frames : List[ndarray]
        list of input images
    classes : List[str], optional
        classes to focus on
    buffer_ratio : float, optional
        proportion of buffer around the width and height of the bounding box
        
    Returns
    -------
    List[dict]
        list of detections for each frame with keys: label, confidence, t, l, b, r, w, h
    

Using sahi_general as a package for inference

• Clone this repository
• Make sure the requirements above are installed
• in the main project folder, install sahi_general as a package

python3 -m pip install --no-cache-dir /path/to/sahi_general

OR as an editable package (if you need to make changes to the code, faster to build)

python3 -m pip install -e /path/to/sahi_general

• import the sahi_general wrapper class for inference (refer to sahi_general_test.py for example usage )

from script.sahi_general import SahiGeneral