Resolved an issue where available disk space was sometimes displayed incorrectly, ensuring clearer and more reliable error messages.
Improvement
Optimized Model Performance
Updated our models for improved processing efficiency and compatibility.
Improvement
Upgraded Redact-Infer
Updated our models with release of new gpu container redact-gpu 3.4.0. Note: This update requires NVIDIA drivers version 555 or higher.
v6.6.0 (11.12.2024)
Type
Description
Details
Bug Fix
Updated Job Processing
Improved messaging to reducing the chances of interruptions and ensuring smoother job processing.
Bug Fix
Increased Stability for Video Processing
Resolved an issue causing jobs to get stuck when using multiple splitters or aggregators, improving stability and efficiency in processing configurations.
Bug Fix
Improved DNAT License Plate Reliability
Fixed inconsistencies in anonymization for license plates, ensuring more dependable results across all video types.
Improvement
Enhanced Security
Addressed and fixed critical vulnerabilities in the Redact pipeline to enhance the security and reliability.
v6.5.0 (12.11.2024)
Type
Description
Details
Improvement
Updated Overlay Output Format
Changed the overlay output format from APNG to TAR files containing PNG frames to improve support for large video files and prevent corruption issues.
Improvement
Clearer Error Messaging
Improved error messages for consumption reporting issues, providing more descriptive feedback to help identify and address problems effectively in case of expired license.
Improvement
Performance Optimisation for Large Files
Enhanced the performance of the video processing pipeline by reducing CPU bottlenecks and improving processing speed.
v6.4.0 (10.09.2024)
Type
Description
Details
Improvement
Security updates
Fixes to known vulnerabilities and dependency version updates.
Improvement
License plate detection model update
Face detection models have been updated with release of new gpu container redact-gpu 3.3.0. Improved the detection accuracy for license plates on trucks and motorcycles.
Bug Fix
Memory Usage for Large Files
Fixed memory usage issues in the Redact pipeline, ensuring smoother and more efficient processing of large video files without overloading system resources.
Bug Fix
Memory Issues for Overlay Processing Large Videos
Resolved out-of-memory issues when processing large video files (over 50 minutes and at least FHD resolution) with overlays.
v6.3.0 (29.07.2024)
Type
Description
Details
Improvement
Faster blur processing for 4K videos
Enhanced the processing speed for high-resolution video frames, ensuring a smoother experience when working with 4K content.
Improvement
Optimized disk space usage
Improved our archive processing to reduce unnecessary disk space usage, resulting in more efficient storage management.
Update
Renamed full body blur parameter
Updated the name of the full body blur parameter for better clarity and ensured no disruption to existing API integrations. The parameter is now called 'full_body_determination_threshold'.
v6.2.0 (02.05.2024)
Type
Description
Details
Improvement
Maintaining archive file names
Users working with archived files can now retain the original names of the files after processing.
v6.1.0 (05.04.2024)
Type
Description
Details
Feature
Added new PII detection
Users with updated licenses now can have access to Full Body Blur (FBB) capabilities in the Precision Blur service & Full body Mask in Mask service. This feature is available with release of new gpu container redact-gpu 3.1.0.
Feature
Added new service (experimental)
Introducing the Mask service, an experimental addition to our service offerings. Mask service for PIIs like full body, faces, and license plates, it blacks out pixels like a censor bar. This feature is available with release of new gpu container Redact-gpu 3.1.0.
Improvement
Redaction quality improvement
Updated tracking service for more accurate and consistent results.
Bug Fix
Bug related to vertical videos
Now that the issue has been resolved, users can process vertical videos (such as those recorded on a camera phone) without experiencing any distortion or unintended alterations.
v6.0.1 (01.02.2024)
Type
Description
Details
Improvement
DNAT improvements
DNAT improvements bringing better face rotation tolerance, face naturalness, and higher processing speed.
Improvement
Security updates
Fixes to known vulnerabilities.
Feature
Extract service retired
Extract service retired due to availability of redaction areas as replacement.
v5.15.1 (24.01.2024)
Type
Description
Details
Improvement
Security updates
Fixes to known vulnerabilities, security improvements, and dependency version updates.
v5.15.0 (30.11.2023)
Type
Description
Details
Bug Fix
Bug related to video splitting
Users will experience improved video splitting without disruptions, ensuring that the output videos maintain the same frame rate as the source material.
v5.14.0 (30.10.2023)
Type
Description
Details
Bug Fix
Addressed a critical issue related to the INSUFFICIENT STORAGE error unit
Users will now experience more user-friendly error messages, allowing for quicker identification and resolution of storage-related issues.
Improvement
Face detection model update
Face detection models have been updated with release of new gpu container redact-gpu 2.7.0.Improved security: aligning with best practices and upgrading the security status.
v5.13.0 (19.09.2023)
Type
Description
Details
Bug Fix
Bug related to the post-processing of detector when using multiple GPUs
the redact-infer container now operates seamlessly within a multiple GPU environment. This means that user can fully leverage the power of multiple GPUs without encountering the previous error , ensuring a smoother and more efficient workflow.
Improvement
Face detection model update
Face detection models have been updated with release of new gpu container redact-infer 2.6.1.This update is designed to enhance detection performance in data domains characterised by low light conditions and the conversion from RAW data format. Please be aware that as a result of these improvements, there may be slight adjustments to the optimal threshold values for non-standard domains.
v5.12.0 (20.07.2023)
Type
Description
Details
Improvement
Significant enhancement to the aggregator service
improve the performance and responsiveness of the aggregator service but also enhance the overall user experience. These enhancements will further streamline your video processing workflows and reduce any delays in frame aggregation specifically for larger videos.
Bug
Fix
Redact Start Time Correction
This correction ensures that speed benchmark results are more accurate and reliable, allowing you to have a better understanding of system performance.
Bug
Fix
The proper generation of AVI. files when processing videos with audio streams
AVI videos with audio streams will now be processed and returned in their correct AVI format. This fix enhances the overall usability of our product and ensures that video processing remains reliable and consistent.
v5.11.2 (14.06.2023)
Type
Description
Details
New feature
Added support for Korean License plates replacements
Deep Natural Anonymisation (DNAT) service now supports replacements for Korean license plates. This feature is only available in API version 4.
Improvement
Face detection model update
Face detection models have been updated with release of new gpu container redact-gpu 2.5.0. This update will help to reduce the False positive rate for specific data sets containing blurry regions. The optimal threshold values for non-standard domains may be slightly altered.
v5.10.0 (04.05.2023)
Type
Description
Details
New feature
Added support for opus audio codec
Audio from videos encoded with the Opus audio codec will be processed unaltered.
New feature
Added filename to the status endpoint
The status endpoint will contain the name of the input file. This feature is only available in API version 4.
New feature
Added Japanese license plates replacement
Deep Natural Anonymization (DNAT) service now supports replacements for Japanese license plates. This feature is only available in API version 4.
Improvement
License plate detection model update
License plate detection models have been updated with release of new gpu container redact-gpu 2.4.0. The optimal threshold values for non-standard domains may be slightly altered.
v5.9.0 (30.03.2023)
Type
Description
Details
Bug Fix
Fixed colors changing after redaction.
Redacted images showed different colors than the source image.
Bug Fix
Aspect ratio of videos taken from phones are changed.
When posting a video from an Iphone the aspect ratio changed.
v5.8.0 (07.03.2023)
Type
Description
Details
Improvement
Improved handling of larger files.
Memory handling in case of large file has been improved to use less memory.
Bug Fix
Fixed a bug where larger files could cause errors.
Uploading large files could cause memory usage increase to the limit and eventually cause an error.
v5.7.0 (09.02.2023)
Type
Description
Details
Bug Fix
Fixed bug where metadata information was lost during processing.
When anonymizing avi files, the metadata would not visible on Windows and the anonymized video could not be played with Windows Media Player. This is now fixed.
v5.6.0 (04.01.2023)
Improved large file support
Improved overall stability
v5.5.0 (06.12.2022)
Added possibility to report consumption through Reverse proxy
Fixed bug with OpenCV that would cause some data to fail
v5.4.1 (16.11.2022)
Correction for archives output to always return frames in RGB colorspace
Improved stability for longer videos
Fixed bug that would cause aspect ratio to change in specific cases
v5.4.0 (31.10.2022)
Resolve the issue of missing last frames when processed as archive for large files
Added filename to the On-Prem UI
v5.3.0 (05.10.2022)
Improve error messaging for redact "heath" endpoint
Faster processing by improved startup time for services
Improved status service with a health REST endpoint
Improved status service to receive new job information
Improved performance for face detector
v5.2.1 (08.09.2022)
Service to export log by customer
Fixed bug to resolve pipeline failure due too many file open
v5.2.0 (31.08.2022)
Added LP confidence final score in the label output
Improved LP detector
v5.1.0 (30.06.2022)
Improve efficiency of tracking service
All occurrences of the env vars are removed
v5.0.0 (18.05.2022)
Removed configurable concurrency
Improve blurring function speed
Service refactoring and consolidation
Deprecate error endpoint
Fixing multiple bugs including issues related to extremely large and small files
v4.4.1-253 (17.02.2022)
Fix deletion of artifacts to avoid disk space issues which cause large files to be stuck
Increase service start time to avoid Kafka re-balancing during start up
Improved LP detector
New micro-service messaging format
v4.2.0-244 (18.10.2021)
Fixed issues with messaging timeouts
Removed race condition in video aggregation
Fixed bounding box errors
Refactored consumption reporting
v4.1.0-243 (20.09.2021)
New retries and timeouts for GRPC clients
Increased processing speed
Better error handling for the death of redact-gpu container
v4.0.0-242 (20.08.2021)
Improved stability
v3.3.1-237 (03.06.2021)
Improved stability
v3.3.1-233 (01.04.2021)
Improved stability
v3.3.0-212 (10.03.2021)
Added support for parallel upload of large files
Improved product usability by removing legacy options person, analytics, block_portraits flag
Renamed IPS_DL_NUM_PROCS to GPU_CONCURRENCY
Renamed IPS_NUM_PROCS to CPU_CONCURRENCY
Fixed race condition in video encoding
v3.2.1-211 (23.02.2021)
Improved stability
v3.2.1-210 (13.02.2021)
Improved UI persistence
Improved input/output consistency of custom labels
Fixed speed issue with custom labels
Fixed handling of spaces in video names
v3.2.0-209 (25.01.2021)
Added support for reprocessing using custom labels
Added persistence to UI
Improved stability
v3.1.1-208 (07.12.2020)
Added experimental support for custom face_determination_threshold
Added quicktime support for HEVC encoded videos
Improved disk usage during processing
Improved scalability of handling incoming http requests
Improved stability
v3.0.0-190 (11.11.2020)
Improved stability
Improved color consistency for synthetic license plate replacements (DNAT)
Added experimental support for custom lp_determination_threshold
v3.0.0-188 (30.10.2020)
Improved bounding box format for metadata endpoint (direct JSON response)
Misc fixes and improvements
v3.0.0-187 (20.10.2020)
Added API endpoint for downloading metadata
Added “Constant Rate Factor” (CRF) parameter for video encoding
Improved focus/usability of API endpoints (removed endpoints for stream outputs and meta data embedding)
