• Added new metrics to the monitoring API. The new metrics can help diagnose poor results caused by dropped frames due to network issues.
• Fixed a bug affecting images saved to disk when the output is either CSV or JSONL. The images were saved, but the channel order used BGR instead of RGB. The issue did not affect webhooks.

• Added vehicle-import API for bulk uploading vehicles.
• On-Premise: app now use Postgresql v14 for data storage. Kindly download the latest docker-compose.yml from the Platerec website and follow the upgrade guide. The data will get migrated automatically to the new version.
• On-Premise: now S3 can be configured as the storage layer.

• Added a new image type that is useful to reduce the bandwidth requirements while still providing context around the license plate.
• Added a new configuration to control the JPEG quality of images sent via webhook.
• Added support for multiple webhook headers in webhooks.

• Added a parameter to customize the camera_id sent via webhook.

• Improved OCR accuracy for the USA, South Africa, Indonesia, Germany, India, Malaysia, the UAE, and Australia.

• Improved make and model classification and coverage.
• Fixed bugs and improved stability and performance.

• Updated the response type for the prediction to "application/json". This matches the response used in Snapshot Cloud.

• Improved vehicle direction accuracy.

• Improved OCR accuracy for the USA, South Africa, Indonesia, Germany, India, Malaysia, the UAE, and Australia.

• Improved make and model classification and coverage.
• Fixed bugs and improved stability and performance.

• Fixed an issue in duplicating Alert Templates
• Web hooks can now be configured to set their acceptable content-type
• Review - now supports updating MCC of visits
• Added visit-detail-api to get individual visit detail by ID.

• Improved users management page
• When system finds similar looking plates, they are shown in Vehicle Merge page.
• Now a site can have more than one Admin user.
• Added region column to Vehicle exports

• Improved OCR accuracy for the USA, Germany, Brazil, Hong Kong, Indonesia, Israel, Argentina, Peru, Malaysia, South Africa, and Colombia.
• Added support for region classification in Malaysia.
• Improved accuracy of the object detection engine.
• Improved accuracy of the make model engine.
• Bug fixes and improvements to stability and performance.

• Improved OCR accuracy for the USA, Germany, Brazil, Hong Kong, Indonesia, Israel, Argentina, Peru, Malaysia, South Africa, and Colombia.
• Added support for region classification in Malaysia.
• Improved accuracy of the object detection engine.
• Improved accuracy of the make model engine.
• Bug fixes and improvements to stability and performance.
• Vehicle orientation information is now enabled by default if you have the MMC option. It is significantly faster and has only a minor impact on overall inference speed.

From 2025-05-08 16:53 GMT+2 until 2025-05-09 11:51 GMT+2, Stream instances configured via our web interface experienced disruptions. The configuration file downloaded from our server was corrupted and prevented Stream from starting. The issue affected instances started during that period. Instances already running Stream were not disrupted.

The issue was caused by the development of a new server-side feature relying on the Stream engine that required changes to the web configuration code. The changes made the configuration file incompatible in some circumstances. Our unit tests failed to catch the issue.

After receiving bug reports informing us of the issue, a fix was quickly implemented and unit tests were put in place to prevent future regressions.

• Alert Templates now support custom Vehicle fields in their message body.
• Now Admins can assign Sites to users to grant access to all cameras under that site.

• A New API to create/update Vehicle comments.
• Alerts now support detection of wrong way driving using the Direction of Travel parameter.
• Various improvements to the Realtime view page and webhooks configuration page.

• Fixed an issue with detection zones not being applied correctly to the image. The masked area was not aligned properly during the object detection process. This issue only impacted users who used detection zones.

• Fixed an issue with detection zones not being applied correctly to the image. The masked area was not aligned properly during the object detection process. This issue only impacted users who used detection zones.

• Improved OCR accuracy for Germany, Peru, Malaysia, Australia, Tajikistan, and the Philippines.
• Added support for region classification for Tajikistan and the Philippines.
• Added precise vehicle orientation information. This feature is only available if you have the MMC option and requires an extra parameter. See the documentation for details.
• Introduced a new object detection engine.

• Improved OCR accuracy for Germany, Peru, Malaysia, Australia, Tajikistan, and the Philippines.
• Added support for region classification for Tajikistan and the Philippines.
• Reduced Docker image size.
• Added progress information when processing video files. See the documentation for more details.
• Introduced a new object detection engine.
• Bug fixes and improvements to stability and performance.