网络输入层data不能支持“python”？

[cuixing158]

最近做“LSTM”时间序列分类，需要用到“python”类型作为输入，发现layer支持的type共有100个，如下 (known types: AbsVal, Accuracy, Arccos, ArgMax, Axpy, BN, BNLL, BatchContrastiveLoss, BatchNorm, BatchReindex, Bias, CenterLoss, ChannelScale, Concat, ConfidencePenaltyLoss, ContrastiveLoss, Convolution, CorrelationLoss, Cosine, Crop, Data, Deconvolution, Dropout, DummyData, ELU, Eltwise, Embed, EuclideanLoss, Exp, FCNAccuracy, Filter, Flatten, Flip, FocalLoss, GeneralContrastiveLoss, GeneralTripletLoss, GlobalLSEMargin, Gram, HDF5Data, HDF5Output, HingeLoss, Hotspot, Im2col, ImageData, InfogainLoss, InnerDistance, InnerProduct, Input, Insanity, LRN, LSTM, LSTMUnit, LabelSpecificAdd, LargeMarginInnerProduct, Local, Log, MVN, MarginInnerProduct, MemoryData, MovingNormalize, MultiLabelImageData, MultinomialLogisticLoss, NCALoss, Noise, Normalize, PReLU, PSROIPooling, Pairwise, Parameter, Permute, Pooling, Power, PredictBox, Proposal, RNN, ROIPooling, ReLU, Reduction, Reshape, Resize, SPP, Scale, Sigmoid, SigmoidCrossEntropyLoss, Silence, Slice, SmoothL1Loss, SoftContrastiveLoss, SoftTruncation, Softmax, SoftmaxWithLoss, Split, SubRegion, TanH, Threshold, Tile, Transformer, TripletLoss, Truncation, WindowData)，没看到支持python的，该如何加入支持呢？

[Trepassing]

Actually,most of caffe support python layer. you need to use the dynamic library of boost to compile caffe.