Release notes for MegaMatcher Accelerator 12

List of releases:

• January 3, 2023
• December 20, 2022
• September 28, 2022
• July 12, 2022
• March 31, 2022
• February 14, 2022
• January 12, 2022
• October 25, 2021
• August 30, 2021
• February 2, 2021
• November 17, 2020
• September 30, 2020

See MegaMatcher Accelerator 13.x release notes for the newer versions.

Algorithms-related changes:

• Updated algorithms version.

Fixes:

• Fixed issue with iris capacity counting for very large systems with Irises Extended license.

Updates and improvements:

• Changed fingers to use large template by default (better accuracy, but requires more RAM).

Fixes:

• OK status is no longer sent if storage is failing to write data.
• Fixed an issue with list IDs when multiple nodes with MMABIS galleries are used.
• Fixed several small bugs and issues.

Updates and improvements:

• ODBC driver updated.

Fixes:

• Fixed several small bugs and issues.

New features:

• New C# and Java samples for gRPC MMA client (found in the Neurotec Biometic SDK zip).
• Introduced lots of algorithm tuning options.

Fixes:

• Fixed several small bugs and issues.

Updates and improvements:

• New version of Docker Compose supported.

Algorithms-related changes:

• Better score normalization when fingerprints are matched without Quality property.

Fixes:

• Fixed issues with incorrect template count calculation in some cases.

Updates and improvements:

• Removed obsolete parameter "fingers use adaptive speed".

Fixes:

• Fixed several small issues.

New features:

• Added new gRPC-based API which removes requirements to use Neurotechnology SDK for communication. Client libraries may be generated and used in practically any programming language.

Fixes:

• Fixed issues where replicated partitions were kept in memory longer than needed.

Updates and improvements:

• Improved performance when a replicated node leaves or joins a cluster.

New features:

• Added backup shell scripts which can be useful for advanced backup scenarios, such as incremental backup.
• Added Irises.SpeedLevel parameter to the webconfig.

Fixes:

• Fixed memory leak issue in some scenarios where lots of delete operations are used.
• Fixed several issues involving network failures when MegaMatcher Accelerator is configured to use both federations and nodes without database.
• Fixed several issues in webconfig Statistics and Federation pages.

Algorithms-related changes:

• Iris engine now features all quality improvements in iris templates extracted using Neurotec Biometric 12.2 SDK.
• Changed default configuration value to convert fingerprint templates to small representation when loading to memory. This results in significantly less RAM usage while not having noticeable impact on accuracy in most cases.

New features:

• Parallel database operations now supported. This can be useful for some ODBC databases which are optimized for multiple concurrent operations.

Fixes:

• Small fixes in webconfig.

Updates and improvements:

• Extended and improved template replication in memory functionality useful for high availability.
• Reduced memory usage in replicated mode.

Fixes:

• Fixed a bug with no database node failing reloading in presence of delete operations.
• Fixed issues in federation mode to properly handle additional connections.

Updates and improvements:

• Loading batch size now is controlled through parameters.

Fixes:

• Fixed the bug which caused failure to insert when palmprints used together with other modalities.
• Fixed several issues in WebConfig UI .

Updates and improvements:

• Increased accelerated faces speed.
• Decreased fingerprint memory usage for previous line templates.

The first release of MegaMatcher Accelerator 12.

Algorithms:

• Fingerprint, face and iris matching engines in the MegaMatcher Accelerator 12.0 Extreme edition now do not require GPU for reaching the highest matching speed. This greatly simplifies deployment without requiring installation of special hardware.
• The fingerprint matching algorithm now provides even higher accuracy, proven by the first place in NIST MINEX III, NIST PFT III and FVC-onGoing fingerprint algorithm evaluations.
• The face matching algorithm now provides higher accuracy.
• The iris matching engine now matches 1.2 Billion irises per second on a single server in the MegaMatcher 12.0 Extreme edition (same speed as for fingerprint and face engines).