Release Notes
Download links:
Nvidia GPU versionwget -O ConPDS_Server_Tracker.AppImage https://ftp.conpds.com/Trackerx86_64_Nvidia | |
Intel GPU versionwget -O ConPDS_Server_Tracker.AppImage https://ftp.conpds.com/Trackerx86_64_Intel |
Only version 0.100
wget -O ConPDS_Server_Tracker.AppImage https://download.conpds.com:/TrackerArm64
Only version 0.100
wget -O ConPDS_Server_Tracker.AppImage https://download.conpds.com:/TrackerArm32
Version 0.133 (Intel + Nvidia GPU's)
- Added more details to logs regarding an issue with get_frame
Version 0.132 (Intel + Nvidia GPU's)
- Added timeout for frame copying (get_frame)
- Added more details to log
Version 0.131 (Intel + Nvidia GPU's)
- Added changes to get_frame (it's expected to get the error and not a picture from get_frame when error)
- Added support of Motion Detection in DNN OCR
Version 0.130 (Intel + Nvidia GPU's)
- Improvement of DNN recognition area to fix cut results
Version 0.129 (Intel + Nvidia GPU's)
- Fixed service start issue on Linux
- Fixed issue with duplicates in camera_group payload
Version 0.128 (Intel + Nvidia GPU's)
- Introduced new MQTT topic:
- camera_group topic
- Fixed performance issue on Intel versio
Version 0.127 (Intel + Nvidia GPU's)
- Introduced new MQTT topics:
- ConPDS/ocr/{server_guid}/groups/{group_uuid}/name
- ConPDS/ocr/{server_guid}/groups/{group_uuid}/number_of_cameras
- Added get_frame_image_data to forward list
Version 0.126 (Intel + Nvidia GPU's)
- Fixed support of Nvidia GPU's on Linux
- Increased the queue for MQTT
Version 0.125 (Intel + Nvidia GPU's)
- Linux DNN OCR Engine for Linux x86_x64 with Nvidia and Intel GPU's.
Version 0.124 (Intel + Nvidia GPU's)
- Fixed crash in streams.
Version 0.123 (Intel + Nvidia GPU's)
- Improved quality of DNN OCR Engine
Version 0.122 (Intel + Nvidia GPU's)
- Improved Recognition model
Version 0.121 (Intel + Nvidia GPU's)
- Added message on the login screen of models compilation waiting
- Added name to result in group
- Fixed order of results in group
- Fixed missed recognition
Version 0.120 (Intel + Nvidia GPU's)
- Added support of Container Detector
- Performance improvements
Version 0.119 (Nvidia GPU's)
- Improved recognition when the server is connected to 2 or more cameras
- Enabled support of TensorRT
- Important! In this version, TensorRT support is enabled, so the server's first launch can take up to 20 minutes. This is due to all 3 models (Container, text detection, and text recognition) being optimised for the GPU where they will be running.
Version 0.118
- Fixed speed of connection while using DNN OCR Engine
- Updated DNN OCR Engine to 2024.04.16
Version 0.117
- Fixed crash in DNN OCR Engine
- Updated Container Detector
Version 0.116
- Fixed crash in DNN engine
- Fix for camera connection when password has % !"#$&'()*+,-./:;<=>?@[\]^_`{|}~
- Added support of ROI for DNN OCR engine
- Added support of Exclude areas in DNN OCR engine
- Added support for Recognition width in the DNN OCR engine
Version 0.115
- Added support of Container Detection to DDN OCR Engine (this is the first version and only for Nvidia for now)
- Based on Container Detection the performance for both OCR models was improved.
- Fixed launch of the server (bug in 0.114)
Version 0.114
- Restored support of Linux x86_64
- Fixed database restoring in the backup feature
- Windows version supports DNN and Classic OCR engines
- Linux supports only Classic OCR engine
Version 0.113
- Fixed crash in DNN OCR Engine
- Added frame saving during engine crash
- The model launch on Intel GPUs was improved
Version 0.112
- Performance improvement for Intel and Nvidia GPUs.
Version 0.111
- Added support for Intel GPUs
Version 0.110
- Updated DNN OCR Engine to 2024.04.9
Improved stability
Improved checksum recognition
Version 0.109
- Fixed a bug when the UI was corrupted at some time, and the page became empty.
- Added backup/restore functionality.
Version 0.108
- Added option Results lifetime. It removes results on the server in Х days.
- Updated FFmpeg from 5.1.4 to 6.1.1 - improved stability and performance.
Version 0.107
- Improved performance and quality of DNN OCR Engine
Version 0.106
- Fixed the server launching with DNN OCR Engine on specific boards.
Version 0.105
- Server: Support of DNN OCR Engine is added.
- To turn on "DNN OCR Engine," go to general settings, check the checkbox for "Enable DNN OCR Engine," confirm turning on DNN OCR Engine, and then click Save.
OCR engine terms:
1. DNN OCR Engine - new CONPDS Core engine based on DNN methods.
2. Classic OCR Engine - old CONPDS Core engine based on classic methods.
- To turn on "DNN OCR Engine," go to general settings, check the checkbox for "Enable DNN OCR Engine," confirm turning on DNN OCR Engine, and then click Save.
Version 0.104
- Server: Improved Intel GPU support
- Server: We have added the ability to republish MQTT topics to a group. The latest version supports topics "results" and "results_two."
Client: Added support of LicensePlate
Client: Added UI for MQTT republishing
License library: Updated network library
License library: Updated OpenSSL to 3.2.1
Version 0.103
- Fixed issue with license
Version 0.100
- Added limit for preview, now if extension is bigger then 1080р, server will decease it
- Improved stability of decoder
Version 0.99
- Fixed empty fields in camera configuration
- Fixed ConPDS Checker Integration
- Improved synchronisation when logging into client
Version 0.98
- Fixed manual licensing
- Support for download from one file
Version 0.96
- Added fix for engine
- Updated model of detector to version 3
- Added auto saving of keys for manual licenses generation
Version 0.94
- Turned off optimisation for preview
Version 0.93
- Fixed problems in UI
- Added the possibility to edit multiple days in scheduler. (Ctrl + click) on day name selects/unselects selection, double click on day unselects all selections
- Added button for cleaning filter/search
Version 0.92
- Fixed platform in license
Version 0.91
- Added scheduler
- Improved vertical container code recognition
- Fixed wrong code recognition (when horizontal container code was recognised, but it's vertical code there)
Version 0.89
- Fixed ocr processing bug introduced in 0.88
- Added tag if checksum not 0
- Improved confidence for NV12
- Fixed crash if pixel format is not supported
Fixed date in history
Fixed pageUp/pageDown
Added preview not only for small icon
Version 0.87
- Added Option to do PgUp and PgDown to move up and down results history
- Performance improvements
- Improved confidence for YUV420 Pixel Format
- Added time stamp to the header of the image preview window
Version 0.86
- Support of coloured codes
Version 0.85
- Added BIC owner to MQTT payload
- Fixed exclude areas
- Added ignoring of not BIC codes
- Fixed editing of ROI
- Fixed memory leak
- Added history loading
- Added filter of results
- Added search of results
- Added code to preview dialog
Version 0.84
- New UI
- Groups logic
- Added page refresh on a new version
Version 0.83
- Fixed OCR cut-out result. Cropped too much.
- Fixed missing MQTT topic on cloned camera
- Fixed missed results when only OCR is enabled
- Fixed removing of old license
Version 0.82
- Fixed crash when a container is not found, and the result is missing.
- Potential fix of memory leak.
Version 0.81
- Fixed launch on Windows Server
- Fixed editing of ROI
- Improved performance of cameras where all features are disabled
Version 0.80
- Fixed reconnecting to MQTT broker in case it was unavailable
- Disabled GPU support in OCR engine for now (because of conflict of cuda implementation with the container detector)
Version 0.79
- Fixed duplicated results
- Improved symbols recognition (by adding contrast)
- Fixed GPU support
Version 0.78
- Fixed duplicates in csv CSV
- Fixed duplicates in Web
- Improved filter for noice in OCR Engine
Version 0.77
- Potential fix for duplicated results
- Improved "Ignore noise after processing"
- Improved stability of web client
Version 0.76
- Fixed missed and wrong results
Version 0.74
- Fixed camera freezing
Version 0.73
- Fixed camera disconnecting
- Added ROI optimization
Version 0.72
- Added the possibility to add exclusion areas. Option Exclude areas will be available only if Container Detection is enabled. In other cases, it's not needed.
Version 0.71
- Added possibility to delete old license
- Fixed Corrupted license.lic
Version 0.70
- Fixed get_frame and do_ocr
Version 0.69
- Added fix for get_frame_image_data without relation to configuration of checker
- Added support of new model for auto licensing
- Added support of new model for manual licensing
Version 0.67
- Added possibility to choose camera features (enable/disable motion, container detector and OCR)
- Added support for new model get_frame
- Added Recognition width, to be able to detect containers close or distant.
- Improved MQTT logging
Version 0.66
- Potential fix for an issue with memory
- Potential fix for an issue with rendering
Version 0.65
- Improved model of detector
- Added support of NMS for improving results of the detector
- Partly fixed issue with long and short containers
- Added limit for frames amount
- Added delay for frames reading
- Added sending results MQTT topic to the camera even if it is in a camera group
- Fixed issue with memory
- Fixed MQTT topic for groups
Version 0.63
- Improved motion detection model
- Improved container detector model
- Added copy camera
- Fixed issue where the version MQTT topic was updated incorrectly
- Fixed memory issue
- Fixed stuck bounding box issue
- Fixed issue with publishing results MQTT if cameras were in a camera group.
Version 0.61
- Added copy camera functionality
- Added Custom Checker Server
Version 0.60
- Added MQTT topic "results" msq.payload to output "UNKNOWN0000" in case of no OCR recognition
- Added support for Container Detection
- Added support for Motion Detection
- Added painting of dynamic ROI
- Added handling for Unknown results
- Added support of get_frame (msg.payload should be container code)
- Added support of arm 64 and arm 32
- Added filter for the size of the object for recognition
- Improved displaying of bounding box
- Fixed ROI
- Fixed motion detection sensitive
- Fixed DepotMaster API url
- Fixed camera heartbeat_timestamp
- Fixed issue with MQTT parser
- Fixed disconnecting of camera on error
- Fixed cropped image of result
- Removed drawing ROI in camera FOV
- Improved detector model
Version 0.59
- Fixed MQTT parser
- Added new "get_frame" MQTT topic
Version 0.58
- Fixed DepotMaster API URL
Version 0.57
- Fixed issue where CSV files were not updated if the camera was in a Camera Group
Version 0.56
- Fixed issue where MQTT topics "uptime" were not updated correctly
Version 0.55
- Fixed crash
- Added DepotMaster server
Version 0.53
- Fixed an issue where some streams didn't show a preview
Version 0.52
- Fixed connection to camera (h.265)
- Fixed OpenCL
Version 0.51
- Fixed launch on windows without cuda
Version 0.50
- Fixed issue with "do_ocr" MQTT topic
Version 0.49
- Fixed stability of streams
Version 0.48
- Fixed Session Timeout logic
- Fixed library name on Windows
Version 0.47
- Added support for Tinker Edge R board
Version 0.46
- Fixed update license
Version 0.45
- Fixed license
Version 0.44
- Improved stability for OpenCL and Cuda
- Fixed hidden groups (MQTT)
- Fixed license renewal issue
- Added retained to next topics: version, uptime, heartbeat_timestamp, configured_cameras and connected_cameras
Version 0.42
- Fixed square slideshow
- Fixed some crashes (not yet on Nano boards)
Version 0.41
- Added showing camera stream as a slideshow instead of a live stream.
- Added option to not include image binaries in the MQTT message.
- Fixed Linux install script, where service started a few seconds after stopping.
- Fixed Camera Group result output
- Fixed Tracker server Online/Offline MQTT topic
- Fixed MQTT messages from non-existing Camera Groups.
- Fixes issue where the app crashed when connecting to H.265 stream.
Version 0.40
- Added GPU processing for Image pre-processing and segmentation (Supports Cuda and OpenCL)
Version 0.37
- Added option for rotating image to 180 degrees
- Fixed mqtt Azena topic
- Added buffers sync (should improve delay for stream)
Version 0.36
- Potential stream fixes (will work if the camera encodes video with h264. Otherwise, a transcoder will be used.HLS needs h264)
Version 0.35:
- Fixed do_ocr
- Fixed OCR rotation
- Added fixes to stream encoding
Version 0.34:
- Fixed problems with stream
- Fixed MQTT LWT topic
- Fixed version sending in MQTT topic
Fixed Server id not unique issue
Fixed Camera uptime MQTT topic
Added (M) to result in client which is result of manual request of MQTT
Added option QOS to Gateway form, default value is 1
Version 0.33:
- Fixed frozen stream
- Fixed timeouts
- Added displaying of backtrace on crash (Linux)
version 0.32:
- Fixed Windows installer
- Fixed MQTT messages
- Decreased video stream delay to 3-4 seconds
Version 0.31:
- Fixed stream delay
- Fixed empty message during install
Version 0.30:
- Added manual licensing
- Added installing AppImage on Linux as a service
Version 0.29:
- Fixed MQTT topics
- Fixed ocr_success
- Fixed ConPDS Checker integration
Version 0.26:
- Fixed do_ocr
- Decreased size of Checker ID
- Fixed file name in Checker
- Added ocr_success to Checker
- Added reconnection to camera after camera editing
- Added automatic name to CVS gateway
- Removed the possibility to edit name for CSV gateway
- Fixed mixup in Camera connection
- Updated LWT message
- Added missed topics for MQTT
- Fixed group results
- Improved work with SQLite
Version 0.25:
- Fixed delay between OCR result and video stream
- Added Last Will Testament to MQTT
- Added Reconnecting status to MQTT
- Fixed reconnecting streams after disconnecting
Version 0.24:
- Fixed connection to mqtt on server start
- Fixed stream issues
Version 0.22:
- Fixed crash on Windows
- Added skipping configuration on server update
- Added Checker Integration
Version 0.20:
- Fixed decoding Asus Tinker Board
- Added errors topic
- Added OCR to MQTT message
- Fixed crash on site(stream could be freezed by it)
Version 0.19:
- Added fixes to stream
- Added fixes to Rockchip MPP
- Improved performance on all platforms
- Added support of Rockchip MPP(hardware decoder for ARM 32)
- Added support of Nvidia Jetson Nano HW Decoder(Nvidia Jetson Nano 2G also should work)
- List of supported Rockchip MPP: RK29XX RK30XX RK31XX RK3288 RK3368 RK3399 RK3228 RK3229 RK3228H RK3328 RV1108 RV1107
Version 0.18:
- Added "iso_size_type" in JSON output
- Fixed CSV mode
- Added "Include Image Binaries in MQTT payload" to Mqtt
- Added MQTT topic to camera dialog
- Changed MQTT topic to "ConPDS/ocr/{server_guid}/
Version 0.17:
- Added new UI for groups
- Improved HLS stream
- Fixed license saving
- Added experimental support for AMD GPU