Version History & Updates

5.8.3Sept 20, 2022

• Minor bug fixes related with colors

5.8.2Sept 15, 2022

• Frontend prediction color fix
• Redis timeout parameters improved
• docker-compose services reordered
• Frontend modal fix for model import
• Custom model import fix for yolov5 models
• Cloud model import problem bypassed

5.8.0Aug 17, 2022

• 'only api mode' added

5.7.0Aug 09, 2022

• Custom model import feature added
• Some bug fixes related to rtsp urls
• Frontend improvements

5.6.0Aug 04, 2022

• Tiling feature added

5.5.5Aug 02, 2022

• Dockerfile changed to fix an encoding problem on alert videos
• New rifle model handler added

5.5.4July 26, 2022

• KafkaConsumer timeout delays increased
• Some css attributes of models and index pages are changed

5.5.3July 21, 2022

• Openvino acceleration activated

5.5.2July 20, 2022

• Frontend fix for better handling portrait videos
• No-glove and weapon models added
• Fix for conf_threshold

5.3.1June 07, 2022

• Klv cpu fixes + analytics endpoint resolved

5.3.0May 27, 2022

• New fall detection model implemented

5.2.9May 26, 2022

• Deepstream version updated to 6.1

5.2.8May 24, 2022

• Klv fix

5.2.7May 20, 2022

• klv output improvement

5.2.6May 19, 2022

• Get original video dimensions update

5.2.5May 16, 2022

• Video upload process upgraded
• File order update
• Local video path changed
• File upload order update
• Kafka removed from upload file process
• Video upload process is now single process for all cameras

5.2.4May 11, 2022

• Triton version is updated

5.2.3May 06, 2022

• Minor bug fixes
• multigpu fix + klv fix + thresholds fix

5.2.1Apr 27, 2022

• application_id + common merge

5.2.0Apr 25, 2022

• Manager fix
• Version update

5.1.6Apr 19, 2022

• min_score and min_nms handler added
• Syntax fix
• Config updated for new version

5.1.5Apr 16, 2022

• Minor bug fixes

5.1.4Apr 14, 2022

• Minor bug fixes

5.1.3Apr 14, 2022

• Versioning feature updates

5.1.2Apr 14, 2022

5.0.0Mar 30, 2022

• Custom model support for ONNX files, tensorflow and pytorch models.
• Model hosting of custom/open source models
• Improved inference engine with robust pipeline and 100% utilization on GPU usage.
• Improved video decoding capabilities with multiple streams and multiple inferences.
• Improved object tracking capabilities.
• Support for various message transmission protocols for big data: MQTT, Kafka.
• Security credentials for Edge dashboard login
• New secure dashboard and backend design powered by React and FastAPI.
• Improved scalability and secure connection options for API usage.
• Ability to deploy Facial and Text models to the Edge.
• Improved facial and object detection models.
• New face recognition module with face-sync capabilities.
• Support for 20+ concurrent streams and individual health checking extension
• New rapid cloud sync feature.
• Improved logging mechanism.
• Enhancements for RTSP streams, youtube videos and m3u live streams.
• New scalable architecture powered by OCI compliant microservices.
• New smart video recording mechanism for generating alert videos.