usama3059 / eq_vision

Make the Data/Model quality evaluation easy for everyone

Easequality

Make the Data and Models Quality easy for everyone

Easequality is basically a High level Wrapper which connects your Data/Model with various Quality Tools, Hence you can evaluate the Data/Model quality with few lines of code.

Updates

• The code will be released soon! Please stay tuned! 😊

Installation

To install

  git clone https://github.com/Usama3059/easequality.git
  cd Easequality

Example/Usage

To run tests, run the following command, we are evaluating on the BCCD Dataset and Yolov5

#Example for Yolov5 Model
from yolo import yolo_quality
cls_dict = {
        0: 'RBC',
        1: 'WBC',
        2: 'Platelets'}


yolo = yolo_quality( train_dir_path='train',
                     test_dir_path='valid',
                     model_path='model.pt',
                     class_dict = cls_dict)

yolo.result()

Integeration

• Supports MSCOCO, PascalVOC, Yolo, Yolo Darknet etc

• Supports Fastai Classification

• Supports Timm Models

• Supports Detectron2

• Supports Yolov5

On going tasks

• Code release for everyone

• Quick Summary

🚀 About Me

Hi, I'm Usama Altaf Zahid mainly work as Computer vision Engineer. I like working with AI models and building Computer vision based problem solving products. This repo is basically a helpful tool to easily and rapidly evaluate and improve the Data/Model Quality, Hence to build more robust Models with very less training process iterations

🔗 Links

Contributing

Contributions are always welcome!

If you have any feedback, please reach out to us at [email protected]