Get reconstructed image and heatmap (no segmentation)

[simon-ami]

I have trained a reverse distillation model on custom images without masks.
how can i predict on an image and retrieve the reconstructed image and the heatmap?
I tried using the inference module with visualization mode on 'full' but it saved only one image with a classification score (like when using 'simple' mode).

Thanks a lot for your amazing contributions!

[simon-ami]

Found a partial answer in discussion [(https://github.com//discussions/819)] :

I had to do the following to extract the heatmap (but still don't know how to get the reconstruction):

from anomalib.deploy.inferencers import TorchInferencer

inferencer = TorchInferencer(config=config, model_source='path/to/model.ckpt')
prediction = inferencer.predict('path/to/image.png')

heat_map = prediction.heat_map
plt.imshow(heat_map)