There is now a pytorch training model, to accommodate the opencv within DNN reasoning (before the rest of the model using opencv reasoning within DNN), will pytorch model into the Torch after Script, use CV: : : within DNN: readNetFromTorch (), error is as follows:
OpenCV (4.2.0) C: \ build \ master_winpack - build - win64 - vc14 \ OpenCV \ modules \ \ SRC \ torch within DNN \ torch_importer CPP: 1022: error: (- 213: The function/feature is not implemented) Unsupported Lua type in function 'CV: : within DNN: : dnn4_v20191202: : TorchImporter: : readObject'

Will turn onnx pytorch model, use the readNetFromONNX (), the error is as follows:
OpenCV (4.2.0) C: \ build \ master_winpack - build - win64 - vc14 \ OpenCV \ modules \ \ SRC \ within DNN within DNN CPP: 562: error: (- 2: Unspecified error) Can 't create layer "82" of type "ConstantOfShape" function in' CV: : : within DNN: dnn4_v20191202: : LayerData: : getLayerInstance '

I am 4.2.0 opencv version, these two mistakes, is a model which does not support of pytorch? How to use the opencv within DNN pytorch model for reasoning,

CodePudding user response:

The opencv within DNN support layer is limited, like biLSTM doesn't support this relatively new layer, layer are commonly used to support,