Profile cover photo
Profile photo
Neurotechnology
113 followers -
Biometric and artificial intelligence technologies and software development kits
Biometric and artificial intelligence technologies and software development kits

113 followers
About
Posts

Post has attachment
MegaMatcher Accelerator Extreme, the fastest biometric engine in the world, is now available. This solution for national-scale projects is capable of matching 1.2 billion fingerprints per second, 1.2 billion faces per second or 700 million eye irises per second on a single server.
Add a comment...

Post has attachment

Post has attachment
Neurotechnology will be exhibiting at ID4Africa 2017, from April 26th to 28th in Windhoek, Namibia. Meet us at booth C12. Contact us to reserve a meeting time.
Add a comment...

Post has attachment
The new SentiVeillance 6.0 improves speed and accuracy for biometric identification using surveillance cameras. The technology includes new face detection and tracking algorithms and support for up to 10 surveillance cameras with one GPU for faster, easier, more accurate identification.
Add a comment...

Post has attachment
Every other week Neurotechnology hosts a meeting to discuss progress in a selected deep learning topic.

This week we will return to recurrent neural networks with a focus on attention mechanisms.

MEETUP: https://www.meetup.com/Deep-Learning-Papers-Review-at-Neurotechnology/

LIVE STREAM: https://youtu.be/hK13NZknUoQ

FACEBOOK: https://www.facebook.com/events/1854672621439121

Take a look at some publications for deeper understanding:
0. Neural machine translation by jointly learning to align and translate. http://arxiv.org/abs/1409.0473
1. Show, attend and tell: Neural image caption generation with visual attention. http://www.jmlr.org/proceedings/papers/v37/xuc15.pdf
2. Dynamic Neural Turing Machine with Soft and Hard Addressing Schemes https://arxiv.org/abs/1607.00036
3. Tracking the World State with Recurrent Entity Networks https://arxiv.org/abs/1612.03969
4. End-to-end memory networks. http://papers.nips.cc/paper/5846-end-to-end-memory-networks
5. Memory Augmented Neural Networks with Wormhole Connections https://arxiv.org/pdf/1409.1259.pdf
6. Machine Learning on Sequential Data Using a Recurrent Weighted Average https://arxiv.org/pdf/1703.01253.pdf

Presentation language: Lithuanian
Add a comment...

Post has attachment

Post has attachment
Add a comment...

Post has attachment
Updated Neurotechnology Biometric and SentiVeillance SDKs.
The revision number of all these SDKs is 156043.

Details on updates:
* Fixed a number of issues with face extraction and template creation.
* Fixed a couple of issues with Face Verification component on Android and iOS.
* Fixed long delay in some scenarios when capturing from RTSP cameras.
* Fixed a number of minor bugs and issues.
* Facial feature points detection now uses a new, smaller data file for the algorithm, which is now more accurate; the new data file should be used now for all face extraction operations.
* Added support for Gradle 3.0 on Android.
* Added support for Suprema BioMini Plus2 and SecuGen Hamster Pro / Pro Duo fingerprint readers.
* Improved segmentation of irises in images with rotated eyes.
* Improved performance of voice template matching.
* Updated iOS samples to support Xcode 8.
* Updated fingerprint scanner integrations updated for Dermalog, Nitgen, Suprema RealScan and SecuGen devices.
* Improved capturing from camera on Android.

Existing customers can download the newest version of the SDKs at the Customers Area.
Add a comment...

Post has attachment
Updated Neurotechnology biometric and SentiVeillance SDKs.
The revision number of all these SDKs is 153687.

Details on updates:
* Fixed video capturing from camera which could cause crash on Windows 10.
* Fixed multiple issues in JavaAbisSample.
* Fixed multiple minor issues in SentiVeillance SDK.
* Added Gradle support for all Java and Android projects (including Android Studio).
* Added reduced size NDF files for medium and large face template generation (recommended for using in mobile devices or applications where file size is crucial). The NDF files with reduced size have a "Lite" suffix in the filenames.
* Added C++ tutorials.
* Added Integrated Biometrics Kojak device support.
* Improved ICAO-compliant face capturing (see Faces or Abis samples).
* Updated JPEG file reading to allow processing more broken files.

Existing customers can download the newest version of the SDKs at the Customers Area. #biometrics #surveillance #SDK  
Add a comment...

Post has attachment
SentiVeillance 5.0 SDK for video surveillance systems with biometric face identification and movement tracking is now available. The new version incorporates the VeriLook 9.0 face recognition algorithm, which was released earlier this year. It provides five times higher accuracy in identifying full frontal faces and 10 to 15 times higher accuracy for unconstrained facial recognition, greatly improving its use for watchlists and other scenarios.
Add a comment...
Wait while more posts are being loaded