how to import lstm net in opencv?
System information (version)
- OpenCV => 4.1
- Operating System / Platform => Windows 64 Bit
- Compiler => Visual Studio 2015
Detailed description:
I want to use LSTM network to classify sequence data in OpenCV, but I have seen many tutorials are image classification. The general way is to use Imread to read the image first, use blobFromImage/blobFromImages to convert the image to blob, then pass the blob. Into the network forward. My problem is that the LSTM network input is not an image, it is a sequence of data, 1024*nImages variable size input data, how to pass into the network for forward?
opencv's dnn can only process pre-trained networks, so you must have a serialized model from some other ml framework (e.g. tf) , no ?
can you elaborate ?
I have trained the LSTM network in matlab and successfully exported it to the onnx model. The LSTM network input is sequence characteristic data of 1024*nImages, and the output is 8 categories.An example of a similar classification problem is here. My model network structure can be seen here: [https://drive.google.com/open?id=1N6S...] But I don't know how to use LSTM network in opencv, that is, I don't know how to input data format to LSTM?