Abstract: Highlights•We focused on recognition scenarios where the testing faces have low-resolutions.•We proposed a multi-scale patch based representation feature learning scheme.•Experimental results demonstrate the effectiveness of our method.
Loading