AutoVu™ SharpZ3 OS 13.7.0 includes important new features as well as updates for bug fixes and security-related issues.
AutoVu MLC improvements
AutoVu MLC™ (Machine Learning Core), which replaces traditional rule-based ALPR technology, has been retrained with the following improvements:
- Vehicle model recognition
- SharpZ3 OS can now identify not only the vehicle make (for example, Honda), but also
the vehicle model (for example, Civic). The Vehicle make analytic has been
renamed Vehicle make and model. When the analytic is enabled, the camera attempts
to identify the make and model and then includes the results in the license plate read
metadata.NOTE: For best results, the street should have adequate lighting. Ensure that the entire vehicle appears in the image and that two sides of the vehicle are visible. As the training of AutoVu MLC continues, you can expect further advancements in SharpZ3 analytics.
- USA ALPR improvements
- The following improvements were made to plate state recognition in the USA:
- Temporary plate recognition
- SharpZ3 OS now recognizes temporary paper license plates that are issued to
vehicles until the official license plates are delivered.NOTE: If the SharpZ3 unit is configured to export license plate reads to an FTP server, you can include temporary license plate metadata in the export.
- Accessible parking plate recognition
- SharpZ3 OS now recognizes the accessible parking license plates that are issued
to drivers with special parking privileges.NOTE: If the SharpZ3 unit is configured to export license plate reads to an FTP server, you can include accessible parking plate metadata in the export.
- Improved plate origin accuracy
- The ability of SharpZ3 OS to identify a plate's state of origin has been improved. If the plate origin cannot be identified, the system outputs an Unknown state name.
- Improved vehicle analytics
- AutoVu MLC has been retrained and now offers improved recognition of vehicle color and vehicle accessories.
General enhancements
- Export plate corner coordinates
- SharpZ3 license plate read metadata now includes the coordinates of the license plate corners, as they appear in the Sharp Portal.
ALPR regional contexts
- New ALPR regional contexts
- The following ALPR regional contexts are now available in the
Analytics page of the Sharp Portal:
- America: General context for countries in North America, Central America, and South America, with the exception of Argentina and Brazil.
- Austria
- Brazil Only: Restricted to Brazil plates only, to reduce false positives.
- Croatia
- Estonia
- Latvia
- LatviaVanity: Reads plates with 3 characters.
- Thailand
- Improved ALPR regional contexts
- The following ALPR regional contexts have been improved:
- Australia
- Improved detection of black plates with white characters. Improved detection of red plates with white characters.
- Denmark
- Improved performance.
- Lebanon
- Now supports plates with Latin characters only.
- Michigan
- Now supports stacked characters.
- North Carolina
- Now supports plates with up to 8 characters.
- Spain
- Improved performance.
- Removed ALPR regional contexts
- The following ALPR regional contexts are no longer supported:
- Japan
- Korea
For more information on available regional contexts, see Supported ALPR contexts.