If you're tired of labeling more data to improve your model, you can use this. Your model labels the objects it detects and writes the coordinates to a txt file. Additionally, it outputs the coordinates on the visual so that you can see if it has correctly identified them. If your model is not detecting some objects well, it means it hasn't labeled them accurately. Therefore, you can manually label the objects that your model struggles with, and let the model handle the rest.