Open xfdu2005 opened 6 years ago
Since don't know what your training data is, i could not give a definite answer. For me, i do cross-modality medical image registration using this method. I just generate two arrays to save file paths for moving images and fixed images respectively. Tensorflow got a filename queue and read image data while executed.
I tried this model for image registration and got a bad result. I suspect that there are some problems in my training set. Could you share how to generate the training data?