d835c2656618fe0d227c7eb15eecb4847e8b6475,lazyflow/operators/obsolete/classifierOperators.py,OpPredictRandomForest,execute,#OpPredictRandomForest#Any#Any#Any#,211

Before Change

        res = self.inputs["Image"][newKey].allocate().wait()

        shape=res.shape
        prod = 1
        for i,e in enumerate(shape):
            if i < len(shape) - 1:
                prod *= e

        features=res.reshape(prod, shape[-1])

        prediction=RF.predictProbabilities(features.astype(numpy.float32))

After Change

        traceLogger.debug("OpPredictRandomForest: Requesting classifier. roi={}".format(roi))
        RF=self.inputs["Classifier"].value
        if RF is None:
            // Training operator may return "None" if there was no data to train with
            result[...] = numpy.zeros(numpy.subtract(roi.stop, roi.start), dtype=numpy.uint8)[...]
            return
        traceLogger.debug("OpPredictRandomForest: Got classifier")        
        //assert RF.labelCount() == nlabels, "ERROR: OpPredictRandomForest, labelCount differs from true labelCount! %r vs. %r" % (RF.labelCount(), nlabels)

        newKey = key[:-1]
        newKey += (slice(0,self.inputs["Image"].shape[-1],None),)

        res = self.inputs["Image"][newKey].allocate().wait()

        shape=res.shape
        prod = numpy.prod(shape[:-1])
        features=res.reshape(prod, shape[-1])

        prediction=RF.predictProbabilities(features.astype(numpy.float32))
        prediction = prediction.reshape(*(shape[:-1] + (RF.labelCount(),)))

        // If our LabelsCount is higher than the number of labels in the training set,
        // then our results aren"t really valid.
        // Duplicate the last label"s predictions
        chanslice = slice(min(key[-1].start, RF.labelCount()-1), min(key[-1].stop, RF.labelCount()))
        result[...]=prediction[...,chanslice]*255 // FIXME: This assumes that channel is the last axis

Instances