I have a plot saved as an image in my

catalog.yml

and the file in the file path exists b/c I ran the pipeline.

scatter_fte_vs_volume:
  type: matplotlib.MatplotlibWriter
  filepath: data/08_reporting/scatter_fte_vs_volume.png
  save_args:
    format: png
    dpi: 300

When I try to load the image in a Jupyter notebook using the catalog using

catalog.datasets.scatter_fte_vs_volume.load()

I get what is probably a hexadecimal number representing the file in some way. Is there a clean Kedro way to load these sorts of images using the catalog, or should I write an explicit

plt.imread

to the

png

file containing the plot?

this datasets does not have a load method. https://github.com/kedro-org/kedro-plugins/blob/7b3ac6c7b88c44143a26fe556475cebb90[…]8/kedro-datasets/kedro_datasets/matplotlib/matplotlib_writer.py

@Erwin That makes sense. Thank you for looking into my question. I'll just load the images directly using matplotlib.

pillow.ImageDataSet or plotly.PlotlyDataSet can be useful

@Erwin Thanks!

Should it support loading?

@Nok Lam Chan I had assumed it should. But I wouldn't claim to know whether it just has not been implemented yet or if there was a design reason why it was not implemented. 🤷‍♂️

def _save(
        self, data: Union[plt.figure, List[plt.figure], Dict[str, plt.figure]]
    )

so should the

_load

method return a

plt.figure

? Feel free to open an issue to discuss this and better if you want to patch this with a PR

yeah this one is tricky, because I would expect a

MatplotlibDataset

to save/load figures , rather than the raw PNG files