c8105b3974fba93097bd28e5ae14a27e5f56c857,ilastik/applets/objectExtraction/objectExtractionSerializer.py,SerialLabelImageSlot,deserialize,#SerialLabelImageSlot#Any#,33

Before Change

        opgroup = group[self.name]
        for inner in opgroup.keys():
            mygroup = opgroup[inner]
            oplabel = innerops[int(inner)]._opLabelImage
            ts = set(numpy.array(mygroup["timesteps"][:]).flat)
            warnings.warn("FIXME: Not serializing label images")
//            assert False, "FIXME: OpLabelImage implementation has changed, no longer has _processedTimeSteps member"
//            oplabel._processedTimeSteps = ts

After Change

        
        mainOperator = self.slot.getRealOperator()
        opgroup = group[self.name]
        for i, (_, subgroup) in enumerate( sorted(opgroup.items() ) ):
            opLabel = mainOperator.getLane(i)._opLabelImage

            for roiString, dataset in subgroup.items():
                logger.debug("Loading labels from dataset: "{}/{}"".format( subgroup.name, dataset.name ))
                // This will be a little slow because the data is passing through a numpy array
                //  instead of somehow directly copying the h5py datasets in their compressed form.
                // We could maybe speed this up, but we"ll lose some abstraction in the cache interface.
                roi = eval(roiString)
                slotRoi = SubRegion( opLabel.Input, *roi )
                opLabel.setInSlot( opLabel.Input, (), slotRoi, dataset[...] )

        self.dirty = False


class SerialObjectFeaturesSlot(SerialSlot):

Instances