Progress report for extract_features

User 4398 | 4/4/2016, 5:32:16 PM

I've been waiting a few hours for this to complete:

# deep_learning_model = graphlab.load_model('http://s3.amazonaws.com/GraphLab-Datasets/deeplearning/imagenet_model_iter45')
image_data['deep_features'] = deep_learning_model.extract_features(image_data)

It would be nice to have an ETA.

Comments

User 91 | 4/4/2016, 9:32:04 PM

On a CPU, this can take about 3 images per second. It should be about 100 times faster on a GPU. Nevertheless, adding the ETA is a great feature request. We hope to include it in a recent release.