Hi,
Is there a way to export a model that uses tf.nn.conv2d_transpose in a fully convolutional network (the input size can change depending on the input image using the network resizing tool)?
Unfortunately, tf.nn.conv2d_transpose requires feeding the output tensor shape as an argument and the way how it's done in tensorflow is to use tf.shape to infer the output shape. However, tf.shape is not supported in SNPE...
Thank you very much in advance,
Kind regards