艾梦

It's totally no problem when using tensorflow backend. Now I test the theano. When running train_model of tutorial.ipynb,we get 1d~2d tensor but not Tensortype(float32,3D) error from T.nnet.softmax() of K.sparse_categorical_crossentropy ```...

Without box.numpy(), nms raise `TypeError: operand 'Tensor' does not support ufuncs (__array_ufunc__=None)`.