CPU usage exceeding 60 GB when running model.predict on single image

[ranti-iitg]

running this
outputs = model.predict_on_batch(np.expand_dims(image, axis=0)),
where model is
resnet50_coco_v0.1.0.h5

on a GPU enabled device consumes more than 60 GB of RAM, after that machine freezes.

1. Device is GPU enabled.
2. Batch size is 1.

[vcarpani]

Can you give us more information: how many classes (and their indices in class files), dimension of the images

[ranti-orbital]

I am just following this notebook with the same image. https://github.com/fizyr/keras-maskrcnn/blob/master/examples/ResNet50MaskRCNN.ipynb

[hgaiser]

There was an inefficiency in the architecture, should be fixed by #34

Could you try that to see if it resolves your issue?

[ranti-iitg]

@hgaiser yup it works, thanks