Skip to content
#

mask-rcnn

Here are 312 public repositories matching this topic...

FrancescoSaverioZuppichini
FrancescoSaverioZuppichini commented Sep 28, 2021

Dear all,

I hope to find you well. The current codebase is totally closed and it is impossible to retrieve all the metrics computed. Allow all metrics to be returned inside dataset.evaluate. For instance, currently, we cannot get the map per class that is printed inside [eval_map](https://github.com/open-mmlab/mmdetection/blob/414c62cb12d1b0ebf6151a697a32b84adc269ca4/mmdet/core/evaluati

FrancescoSaverioZuppichini
FrancescoSaverioZuppichini commented Oct 12, 2021

Currently, it seems impossible to just get a list of bounding boxes from the predictions.

img = np.array(Image.open("..."))
infer_ds = Dataset.from_images([img], valid_tfms, class_map=dm.parser.class_map)

preds = model_type.predict(model, infer_ds, keep_images=True)
# I just want the bounding boxes 

preds is a list of Prediction and the codebase is not easy to navi

Improve this page

Add a description, image, and links to the mask-rcnn topic page so that developers can more easily learn about it.

Curate this topic

Add this topic to your repo

To associate your repository with the mask-rcnn topic, visit your repo's landing page and select "manage topics."

Learn more