Todo: 详细描述模型及OP支持情况, 包括不同加速平台的支持情况以及包括OP在不同框架间的对应关系
目前 TNN 支持常用的 CNN 的网络：
- Classical CNN: Vgg AlexNet GoogleNet(v1,v2,v3)
- Practical CNN: ResNet DenseNet SENet
- Light-weight CNN: SqueezeNet MobileNet(v1,v2,v3) ShuffleNet(v1,v2) MNasNet
- Detection: Mtcnn-v2
- Detection: Vgg-ssd SqueezeNet-ssd MobileNetv2-SSDLite ...
- Detection: Yolo-v2 MobileNet-YOLOV3 ...
- Segmentation: FCN PSPNet
| model name | onnx2tnn | Naive | armv7 | armv8 | opencl | metal |
|---|---|---|---|---|---|---|
| AlexNet | yes | yes | ||||
| DenseNet(121) | yes | yes | ||||
| FCN | Yes | yes | ||||
| GoogleNet-v1 | yes | yes | ||||
| GoogleNet-v2 | yes | yes | ||||
| GoogleNet-v3(inception) | yes | yes | ||||
| MnasNet | yes | yes | ||||
| MobileNet-v1-ssd(caffe) | yes | yes | - | - | - | - |
| MobileNet-v1-ssd(pytorch) | yes | yes | ||||
| MobileNet-v2-SSDLite | yes | yes | ||||
| MobileNet-yolov3 | ? | ? | ||||
| MobileNet-v1 | yes | yes | ||||
| MobileNet-v2 | yes | yes | ||||
| MobileNet-v3(small,large) | yes | yes | ||||
| Mtcnn-v2 | yes | yes | ||||
| PSPNet | yes | yes | ||||
| ResNet50 | yes | yes | ||||
| SENet(154) | yes | yes | ||||
| ShuffleNet-v1 | yes | yes | ||||
| ShuffleNet-v2 | yes | yes | ||||
| SqueezeNet-ssd | No | - | - | - | - | - |
| SqueezeNet-v1 | yes | yes | ||||
| UNet | yes | yes | ||||
| Vgg-ssd | yes | yes | ||||
| Vgg16 | yes | yes | ||||
| Yolo-v3-tiny | yes | yes | ||||
| Yolo-v2 | ? | ? | ||||
| Yolo-v2-tiny | yes | yes | ||||
| Yolo-v3 | yes | yes |
- 关于 upsample 的计算,当参数mode == "bilinear" 或者 mode == "linear", pytorch 转化出的 onnx 模型是有问题的，pytorch 和 onnx 的计算结果是不对齐的。这是 onnx 本身的 bug，这一点尤其需要注意。但是遇到这种情况请不要担心，将转换后的 ONNX 模型转换为 TNN 后，我们保证了 TNN 和 Pytorch 的计算结果是对齐的。经过测试发现会出现上述问题的网络模型有 FCN 以及 PSPNet。
- 上述表格中的 "?" 表示未知，由于暂时找不到相应类型的模型，所以暂时无法就该模型的兼容性进行测试。
| TNN Operators | ONNX Operators | Naive | armv7 | armv8 | opencl | metal |
|---|---|---|---|---|---|---|
| Abs | Abs | yes | yes | yes | yes | yes |
| Acos | Acos | yes | yes | yes | ||
| Add | Add | yes | yes | yes | yes | yes |
| Asin | Asin | yes | yes | yes | ||
| Atan | Atan | yes | yes | yes | ||
| BatchNormCxx | BatchNormalization | yes | yes | yes | yes | yes |
| Cast | Cast | |||||
| Ceil | Ceil | yes | yes | |||
| Clip | Clip | yes | yes | yes | yes | yes |
| Concat | Concat | yes | yes | yes | yes | yes |
| Const | Constant | |||||
| Convolution | Conv | yes | yes | yes | yes | yes |
| Convolution(depthwise) | Conv | yes | yes | yes | yes | yes |
| Convolution(group) | Conv | yes | yes | yes | yes | yes |
| Cos | Cos | yes | yes | yes | ||
| Deconvolution | ConvTranspose | yes | yes | yes | yes | yes |
| Deconvolution(depthwise) | ConvTranspose | yes | yes | yes | yes | yes |
| Deconvolution(group) | ConvTranspose | yes | yes | yes | yes | yes |
| DepthToSpace | DepthToSpace | |||||
| DetectionOutput | DectectionOutput(custom operator) | yes | ||||
| Div | Div | yes | yes | yes | yes | yes |
| Dropout | Dropout | |||||
| Elu | Elu | yes | yes | yes | yes | yes |
| Exp | Exp | yes | yes | yes | yes | yes |
| Flatten | Flatten | |||||
| Flatten | Shape+Gather+Constant+Unsqueeze+Concat+Reshape | |||||
| Floor | Floor | yes | yes | yes | ||
| Gather | Gather | |||||
| HardSigmoid | HardSigmoid | yes | yes | yes | yes | yes |
| HardSwish | Add + Clip + Div + Mul | yes | yes | yes | yes | yes |
| HardSwish | Add + Clip + Mul + Div | yes | yes | yes | yes | yes |
| HardSwish | HardSigmoid + Mul | yes | yes | yes | yes | yes |
| InnerProduct | Gemm | yes | yes | yes | yes | yes |
| InstBatchNormCxx | InstanceNormalization | yes | yes | yes | yes | yes |
| LRN | LRN | yes | ||||
| Log | Log | yes | yes | yes | ||
| LogSigmoid | Sigmoid + Log | yes | yes | yes | yes | yes |
| MatMul | Matmul | |||||
| Max | Max | yes | yes | yes | yes | yes |
| Min | Min | yes | yes | yes | yes | yes |
| Mul | Mul | yes | yes | yes | yes | yes |
| Neg | Neg | yes | yes | yes | yes | |
| Normalize | ReduceL2+Clip+Shape+Expand+Div | yes | yes | yes | yes | yes |
| Normalize | Reduce + Clip + Expand + Div | yes | yes | yes | yes | yes |
| Normalize | Mul(square)+Reduce+Max+Sqrt+Mul | yes | yes | yes | yes | yes |
| PRelu | LeakyRelu / PRelu | yes | yes | yes | yes | yes |
| Pad | Pad | yes | yes | yes | yes | yes |
| Permute | Transpose | yes | yes | yes | yes | |
| Pooling (Avg) | AveragePool | yes | yes | yes | yes | yes |
| Pooling (GlobalAverage) | GlobalAveragePool | yes | yes | yes | yes | yes |
| Pooling (GlobalMax) | GlobalMaxPool | yes | yes | yes | yes | yes |
| Pooling (Max) | MaxPool | yes | yes | yes | yes | yes |
| Power | Pow | yes | yes | yes | yes | yes |
| PriorBox | PriorBox(custom operator) | yes | yes | |||
| Reciprocal | Reciprocal | yes | yes | yes | ||
| ReduceL1 | ReduceL1 | yes | ||||
| ReduceL2 | ReduceL2 | yes | yes | |||
| ReduceLogSum | ReduceLogSum | yes | yes | |||
| ReduceLogSumExp | ReduceLogSumExp | yes | yes | |||
| ReduceMax | ReduceMax | yes | yes | yes | ||
| ReduceMean | ReduceMean | yes | yes | yes | yes | yes |
| ReduceMin | ReduceMin | yes | yes | |||
| ReduceProd | ReduceProd | yes | yes | |||
| ReduceSum | ReduceSum | yes | yes | yes | yes | yes |
| ReduceSumSquare | ReduceSumSquare | yes | yes | |||
| Relu | Relu | yes | yes | yes | yes | yes |
| Relu6 | Clip | yes | yes | yes | yes | yes |
| Reorg | DepthToSpace | yes | yes | |||
| Repeat | Tile | |||||
| Reshape | Reshape | yes | yes | yes | yes | yes |
| Selu | Selu | yes | yes | yes | ||
| ShuffleChannel | Reshape + Transpose + Reshape | yes | yes | yes | yes | yes |
| Sigmoid | Sigmoid | yes | yes | yes | yes | yes |
| Sign | Sign | yes | yes | yes | yes | yes |
| SignedMul | Sub+Sign+Add+Div+Gather+Slice+Mul | |||||
| Sin | Sin | yes | yes | yes | ||
| Slice(StrideSlice) | Slice | yes | yes | yes | yes | yes |
| Softmax | Softmax | yes | yes | yes | yes | yes |
| Softplus | Softplus | yes | yes | yes | yes | yes |
| Split | Split | yes | ||||
| Sqrt | Sqrt | yes | yes | yes | yes | yes |
| Squeeze | Squeeze | |||||
| Sub | Sub | yes | yes | yes | yes | yes |
| Sum | ||||||
| Tan | Tan | yes | yes | yes | ||
| Tanh | Tanh | yes | yes | yes | yes | yes |
| Unsqueeze | Unsqueeze | no | ||||
| Upsample | Upsample / Resize | yes | yes | yes | yes | yes |
- 在上面的表格中 TNN 的HardSwish 算子对应 ONNX 算子是"Add + Clip + Div + Mul"，这代表 ONNX 的四个算子共同组合成了 TNN 中的 HardSwsh 算子。表格中的 "+" 符号代表算子的组合，其他类似的算子同理。
- 在上面的表格中 TNN 的 PRelu 算子对应着 ONNX 的"LeakyRelu / PRelu"，这代表 TNN 中的 RRelu 算子同时支持 ONNX 中的 PRelu 和 LeakyRelu 算子。
| device | support |
|---|---|
| ARMv7 | Yes |
| ARMv8 | Yes |
| OpenCL | Yes |
| Metal | Yes |