ValueError: No op named NonMaxSuppressionV2 in defined operations

[tonmoyborah]

I have downloaded the models, downloaded and compiled protoc 3.4 and I have tensorflow-gpu 1.1 and tf cpu 1.2 (in an anaconda env). I am working on windows 10. When I run the code in object-detection-tutorial the code throws this error in "Load a (frozen) Tensorflow model into memory" part.

Any particular reason? Is this because of old tf version? Anaconda doesn't have newer versions for windows, so I am hoping the version isn't the case.

[derekjchow]

I believe this is due to old tf version. Could you try upgrading to version 1.4?

[tonmoyborah]

I think it is. I couldn't test it but I have seen it working on tf 1.4

[drimyus]

Hi, I have the same error on tensorflow object detection model with tensorflow-cpu-1.8.1.
No op named NonMaxSuppressionV2 in defined operations.
model : SSD_Mobile, tensorflow version tf-(cpu/gpu) 1.8.1