Operator lists

Operator Supported Remark
AVERAGE_POOL_2D Y  
ARGMAX Y Only CPU and TensorFlow is supported.
BATCH_NORM Y Fusion with activation is supported.
BATCH_TO_SPACE_ND Y  
BIAS_ADD Y  
CAST Y Only CPU and TensorFlow model is supported.
CHANNEL_SHUFFLE Y  
CONCATENATION Y For GPU only support channel axis concatenation.
CONV_2D Y Fusion with BN and activation layer is supported.
CROP Y Only Caffe's crop layer is supported (in GPU, offset on channel-dim should be dividable by 4).
DECONV_2D Y Supports Caffe's Deconvolution and TensorFlow's tf.layers.conv2d_transpose.
DEPTHWISE_DECONV2D Y Supports Caffe's Group and Depthwise Deconvolution. For GPU only supports depthwise mode(multiplier = 1).
DEPTHWISE_CONV_2D Y Only multiplier = 1 is supported; Fusion is supported.
DEPTH_TO_SPACE Y  
DEQUANTIZE Y Model quantization will be supported later.
ELEMENT_WISE Y ADD/MUL/DIV/MIN/MAX/NEG/ABS/SQR_DIFF/POW/RSQRT/SQRT/EQUAL/FLOOR_DIV
ELU Y  
EMBEDDING_LOOKUP Y  
EXPANDDIMS Y Only CPU and TensorFlow is supported.
FILL Y Only CPU and TensorFlow is supported.
FLATTEN Y Only Caffe is supported.
FULLY_CONNECTED Y  
GROUP_CONV_2D   Caffe model with group count = channel count is supported.
IDENTITY Y Only TensorFlow model is supported.
LOCAL_RESPONSE_NORMALIZATION Y  
LOGISTIC Y  
LSTM    
MATMUL Y Only CPU is supported.
MAX_POOL_2D Y  
ONE_HOT Y Only TensorFlow model is supported.
PAD Y  
PSROI_ALIGN Y  
PRELU Y Only Caffe model is supported
PRIOR_BOX Y Only Caffe model is supported
REDUCE_MEAN Y Only TensorFlow model is supported. For GPU only H + W axis reduce is supported.
RELU Y  
RELU1 Y  
RELU6 Y  
RELUX Y  
RESHAPE Y Limited support: GPU only supports softmax-like usage, CPU only supports the usage which not change the storage format.
RESIZE_BICUBIC Y Only Tensorflow is supported
RESIZE_BILINEAR Y Only Tensorflow is supported
RESIZE_NEAREST_NEIGHBOR Y Only Tensorflow is supported
REVERSE Y Only CPU and Tensorflow is supported
RNN    
RPN_PROPOSAL_LAYER Y  
SHAPE Y Only CPU and TensorFlow is supported.
STACK Y Only CPU and TensorFlow is supported.
STRIDEDSLICE Y Only CPU and TensorFlow is supported.
SPLIT Y In Caffe, this op is equivalent to SLICE; For GPU only support channel axis slice.
SOFTMAX Y  
SPACE_TO_BATCH_ND Y  
SPACE_TO_DEPTH Y  
SQUEEZE Y Only CPU and TensorFlow is supported.
TANH Y  
TRANSPOSE Y Only CPU and TensorFlow is supported.
UNSTACK Y Only CPU and TensorFlow is supported.