DeepSign: Deep On-Line Signature Verification

Abstract

Deep learning has become a breathtaking technology in the last years, overcoming traditional handcrafted approaches and even humans for many different tasks. However, in some tasks, such as the verification of handwritten signatures, the amount of publicly available data is scarce, what makes difficult to test the real limits of deep learning. In addition to the lack of public data, it is not easy to evaluate the improvements of novel proposed approaches as different databases and experimental protocols are usually considered.The main contributions of this study are: i) we provide an indepth analysis of state-of-the-art deep learning approaches for on-line signature verification, ii) we present and describe the new DeepSignDB on-line handwritten signature biometric public database 1 , iii) we propose a standard experimental protocol and benchmark to be used for the research community in order to perform a fair comparison of novel approaches with the state of the art, and iv) we adapt and evaluate our recent deep learning approach named Time-Aligned Recurrent Neural Networks (TA-RNNs) 2 for the task of on-line handwritten signature verification. This approach combines the potential of Dynamic Time Warping and Recurrent Neural Networks to train more robust systems against forgeries. Our proposed TA-RNN system outperforms the state of the art, achieving results even below 2.0% EER when considering skilled forgery impostors and just one training signature per user.

Dynamic Handwritten Signature Identification Using Spiking Neural Network

“…The MCYT-330 database [17], which is a part of DeepSignDB [5,20], was chosen for the experimental study of the proposed DSI system performance. For this purpose, it would be necessary to construct a spiking neural network on Fig.…”

“…The experimental protocol proposed in [5,20] does not provide for the use of skilled forged signatures to train the DSI system. And our proposed DSI system provides such an opportunity for training.…”

“…Such secondary parameters suggest taking the speed of change of coordinates (v x = dX/dt, v y = dY/dt), the acceleration of change of coordinates (a x = dv x /dt, a y = dv y /dt), as well as various discrete features, such as the number of maxima, minima, convex and concave areas, etc. [1,3,20]. Not all dynamic parameters are taken at the same time, but certain of their optimal sets.…”

“…Any database of user signatures can be used to train the system, in which the primary dynamic parameters of signatures (X(t), Y(t) and P(t)) are stored. DeepSignDB [5,20] was used in this study.…”

AirSign: Smartphone Authentication by Signing in the Air

“…FaceID [8] TrueDepth camera Large 3D head mask attack [26] >99.9% Samsung FR [27] RGB camera Medium Images attack [18] -TouchID [5] Fingerprint sensor Large Finger masks [20] >99.9% Samsung FP [6] Ultrasonic fingerprint sensor None Finger masks [19] >99.9% PIN [3] Smartphone screen None Shoulder-surfing attack [4] -AirAuth [21] Depth camera Large Additional hardware EER 3.4% Z. Sitová et al [23] Motion sensors and Large Low accuracy EER 7.16% (walking) touch screen EER 10.05% (Sitting) EchoPrint [22] Acoustic sensors and camera Medium Low accuracy in low illumination 93.5% SilentSign [28] Acoustic sensors Small Handwritten signature by pen EER 1.25% ASSV [25] Acoustic sensors Small Handwritten signature by pen EER 5.5%…”