How to show 3D CNN saliency map

[cloud7878]

Hello!
I met a question when i found saliency map in 3D CNN, the question is as follows:
net = caffe.Classifier(CNN_ARCH_FILE, PRETRAINED_MODEL,
mean_file=np.load(MEAN_FILE).mean(1).mean(1),
channel_swap=(2,1,0),
input_scale=255,
image_dims=(16,150,100))

ValueError Traceback (most recent call last)
in ()
22 channel_swap=(2,1,0),
23 input_scale=255,
---> 24 image_dims=(16,150,100))

/home/liuyun/C3D-caffe/caffe/python/caffe/classifier.pyc in init(self, model_file, pretrained_file, image_dims, gpu, mean_file, input_scale, channel_swap)
31 self.set_mode_cpu()
32
---> 33 if mean_file:
34 self.set_mean(self.inputs[0], mean_file)
35 if input_scale:

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()