train 'mobilenet_v2' with data_format 'NCHW' , is it right here?

[zymazloa]

https://github.com/tensorflow/benchmarks/blob/6c2ccb45049673f09fdea9406372d6561db5c4fd/scripts/tf_cnn_benchmarks/models/mobilenet_v2.py#L47

does it mean slim.conv2d use default data_format NHWC here ?

[reedwm]

/CC @mingxingtan, can you comment here?

[BunnyShan]

@zymazloa mobilenet need to use NHWC, and if you use synthetic_gpu_images, the data_format must be set to NHWC. Also I notice mobilenet_v2 training speed is about 250images/sec with the right data_format, I guess mobilenet may be not suitable to train on GPU devices.