Improving Hand Recognition in Uncontrolled and Uncooperative Environments using Multiple Spatial Transformers and Loss Functions
preprint paper can be found on arXiv here
you can ask for some Python (pytorch) codes and quater million marked hand key points (see Fig. 3 in paper) by email maskotky@gmail.com
To acquire the NTU Palmprints from the Internet (NTU-PI-v1) database (which was used in the paper), download and fill in the "Data Release Agreement.pdf" file. Print the agreement and sign on page 2. Scan the signed copy and send back to maskotky@gmail.com, and adamskong@ntu.edu.sg with title "Application for NTU Palmprints from the Internet database (NTU-PI-v1)". A download link to the database will be send to you once after we receive the signed agreement file.