Currently, only metadata is sent to the cloud, but not the images. Idea would be to introduce a device method that could be called from a cloud component so that images can be processed there.
One use case would be to request images that have low prediction accuracy to the cloud to be re-labelled and used while re-training the model.
Currently, only metadata is sent to the cloud, but not the images. Idea would be to introduce a device method that could be called from a cloud component so that images can be processed there.
One use case would be to request images that have low prediction accuracy to the cloud to be re-labelled and used while re-training the model.