Beyond “Near-Duplicates”: Learning Hash Codes for Efficient Similar-Image Retrieval

   Abstract