Caffe: Can caffe handle inputs as matrixes in float type?

[yesafdsaf]

I am using caffe to do a classification task with every example is a float matrix, I am wondering whether there is a specific data layer which is able to handle such kind of inputs?

[n-zhang]

You can put any type of float data in leveldb or lmdb and it doesn't have to be images. You can refer convert_imageset.cpp and modify based on that to create leveldb for your matrix.

We welcome you to look at the extensive documentation, tutorials, and examples at http://caffe.berkeleyvision.org/
However, please ask usage questions on caffe-users -- this issues tracker is primarily for Caffe development discussion. Thanks!

[yesafdsaf]

@n-zhang Thank you very much!

[xulifan]

To write data into lmdb:
if vector.dtype == np.int:
datum.data = vector.tostring()
elif vector.dtype == np.float:
datum.float_data.extend(vector.flat)

To read int data in lmdb:
flat_x = np.fromstring(datum.data, dtype=np.int)
x = flat_x.reshape(datum.channels, datum.height, datum.width)
y = datum.label

To read float data in lmdb:
x = np.array(datum.float_data).astype(float).reshape( datum.channels, datum.height, datum.width)
y = datum.label

[ghost]

@xulifan Thank you !