New Callback to record scores.

[garland3]

I made a new callback to output the scores to csv on inference. I'm new to this project, so I'm not sure if you accept new callbacks or what???? Just wanted to find out before I make a PR.

in get_callbacks I have added.

# add my ScoreOutput call back
    if  hasattr( config.project, "output_scores") and config.project.output_scores==True:
        write_score = WriteScoreCallBack(config.visualization.image_save_path)
        callbacks.append(write_score)

and then my actual callback is .

"""Write Score Callback."""
from typing import Any
from pathlib import Path
import pandas as pd
import pytorch_lightning as pl
from pytorch_lightning.utilities.types import STEP_OUTPUT
from pytorch_lightning.callbacks import Callback

class WriteScoreCallBack(Callback):
    def __init__(self, save_path):
        super().__init__()  # Call the parent's initializer
        if save_path is not None:
            self.save_path = Path(save_path)  # Save_path is a Path object

    def on_predict_batch_end(
        self,
        trainer: pl.Trainer,
        pl_module: pl.LightningModule,  # Use the base class 
        outputs: dict | None,  # Outputs from the predict step are typically dicts
        batch: Any,
        batch_idx: int,
        dataloader_idx: int,
    ) -> None:

        # Check that outputs is not None and contains the expected keys
        if outputs is not None and "pred_scores" in outputs and "image_path" in outputs:
            csv_name = self.save_path / "scores.csv"
            
            # Check if the CSV file already exists
            if csv_name.exists():
                df = pd.read_csv(csv_name)
            else:
                df = pd.DataFrame(columns = ['pred_scores', 'image_path'])

            pred_scores = outputs['pred_scores']
            pred_scores = pred_scores.cpu()
            filenames = [str(Path(a)) for a in  outputs["image_path"]]  # Convert Paths to strings for DataFrame
            df_new = pd.DataFrame({'pred_scores': pred_scores, 'image_path': filenames})

            df_combined = pd.concat([df, df_new])
            df_combined.to_csv(csv_name, index=False)
    ```