Dear Corvid, Zabeth,
Hard to say but NC means "Batch Size, Number of Channels".
OpenVIno Inference Engine deep learning layer format is a 4-dimensional shape in the format [N,C,H,W] Where N = Batch Size, C = Num of Channels, H = Height and Width = Width. Tensorflow is different, it goes with [N,H,W,C]. In the case of Tensorflow you'd pass in input shape exactly like [N,H,W,C] but Inference Engine will perform a conversion to NCHW.
Size of dims(4) and format(NC) are inconsistent.
Is telling me though that somewhere Inference Engine is seeing a dim size of 4 but yet only getting NC (a dim size of 2). Without seeing your code and your model it's difficult to diagnose.
Let me know more and I'll be glad to help.