Skip to content

Latest commit

 

History

History
180 lines (140 loc) · 7.32 KB

CHANGELOG.md

File metadata and controls

180 lines (140 loc) · 7.32 KB
Raw

Change Log

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[4.0.2] - 2024-05-04

Added

  • added Paris (eu-west-3) and Sydney (ap-southeast-2) regions where Anthropic Claude 3 Haiku are available

[4.0.1] - 2024-04-11

Bugfixes

  • AWS Elemental MediaConvert does not create frameCapture group when s3://[PROXY_BUCKET]/_settings/aioption.json is missing.
  • Short form video (5s) fails the analysis.
  • Rephrase the Version Compatibility input parameter on the CFN template to be more clear.

New features

  • added Top 5 most relevant tags at the scene level

[4.0.0] - 2024-03-06

New features

  • Dynamic frame analysis workflow
  • Auto face indexing workflow
  • Scene and Ad break detection workflow
  • GenAI playground with Amazon Bedrock (Anthropic Claude)
  • Knowledge graph with Amazon Neptune Serverless
  • Option to choose Amazon OpenSearch Serverless Service instead of Amazon OpenSearch Service (cluster)
  • Complex search query that supports AND, OR, NOT directives.
  • Experimental feature: Shoppable Metadata (Disabled by default). Contact your AWS representative if you are interested in this feature.

Changes

  • Updated lambda function runtime to NodeJS 20.x
  • Updated AWS SDK JS to version 3
  • Refactored video analysis state machine by introducing new sub-state machines
    • Dynamic frame segmentation state machine
    • Video based detection state machine
    • Frame based detection state machine
    • Custom model detection state machine
    • Analysis Post Process state machine
  • Consolidated opensearch indices into a single index to support complex search query

Added

  • An unified state machine status event bus using Amazon EventBridge to consolidate events from various state machines
  • A new Amazon DynamoDB table, faceindexer to store face id and name mapping
  • An Update Face Indexer state machine to manage the face indexer logics such as adding, modifying, and removing faces from the face indexer table, opensearch documents, and metadata files on proxy bucket
  • A Graph Indexer state machine to manage the Amazon Neptune Serverless logics such as adding, modifying, and removing nodes and relationships from the graph database
  • Amazon CodeBuild to build and package opensource models into docker images and store in a private Amazon Elastic Container Registry (Amazon ECR)
    • CLIP, opensource zero shot image classification model running in containerized lambda function for generating image embeddings, used for scene detection feature
    • Faiss, opensource vector store running in containerized lambda funciton for similarity search (stateless), used for scene detection feature
    • OWL-ViT, opensource zero-shot object detection model running in containerized lambda function for apparel detection, used for shoppale metadata feature
    • BLIP, opensource text to caption model running in containerized lambda function for generating image caption, used for image analysis
  • Private VPC for the Amazon Neptune Serverless instance

Removed

  • Amazon Rekognition Person Pathing API

[3.1.5] - 2023-11-02

Security

  • Security updates

Added

  • solution-manifest.yaml file
  • package-lock.json files

[3.1.4] - 2023-09-05

Changed

  • Updated backend to Nodejs 16

Fixed

[3.1.3] - 2023-08-01

  • Fixed an issue where media analysis result is not visible in the web application

[3.1.2] - 2023-04-20

  • Updated object ownership configuration on the S3 buckets.
  • Deploying previous versions of the solution will fail due to Amazon S3 security changes. More info on this blog

[3.1.1] - 2023-04-3

Added

  • Added package-lock.json files to all lambda packages.

[3.1.0] - 2023-02-28

Added

  • AppRegistry integration
  • SonarQube properties file

Changed

  • Code fixes for SonarQube
  • CloudFormation Stack Update is not supported when upgrading from v3.0.0 to v3.1.0. A new stack must be deployed.

Contributors

  • @sandimciin
  • @eggoynes

[3.0.0] - 2022-02-01

Added

  • Amazon Rekognition Custom Labels model support
  • Amazon Rekognition Segment API support
  • Amazon Rekognition Text detection for video
  • Amazon Transcribe Custom Vocabulary and Custom Language Model support
  • Amazon Comprehend Custom Entity Recognizer support
  • Support audio and document files
  • Frame based analysis
  • Backlog queuing mangagement system
  • Rewrite backend state machines to use Nested state machine and Map state to simplify workflow
  • Enable AWS-XRAY tracing on all AWS Lambda functions

Changed

  • Updated search engine to Amazon OpenSearch 1.0
  • Redesigned UI/UX to enable different type of collections
  • Advanced search feature to pinpoint timestamps of the search results

Removed

  • Removed Amazon SageMaker Ground Truth Labeling workflow for face indexing

[2.0.1] - 2019-12-13

Added

  • added additional Comprehend and Transcribe languages support to Media2Cloud

Changed

  • workaround audio analysis issue when Comprehend result contains leading punctuation characters.

Removed

[2.0.0] - 2019-10-17

Added

  • Backend: new analysis engine with analysis-monitor, audio-analysis, video-analysis, image-analysis state machines
  • Backend: Image support (JPG, PNG, BMP, and selective RAW images), ability to extract EXIF, GPS, XMP information
  • Backend: enhance analysis detection
  • Backend: server side md5 checksum
  • Backend: auto restore object from GLACIER, DEEP_ARCHIVE storage
  • AIML: Celebrity detection
  • AIML: Face Matching with your own face collection
  • AIML: Label detection
  • AIML: Moderation detection
  • AIML: Person pathing
  • AIML: Text detection on image
  • AIML: Transcription (subtitle) generation
  • AIML: Entity detection
  • AIML: Keyphrase detection
  • AIML: Sentiment detection
  • AIML: Topic detection
  • UI: support cropping image and index faces to Amazon Rekognition Face Collection
  • UI: Ability to index face and store to your own face collection
  • UI: Ability to adjust the confidence level of the detection
  • UI: Ability to enable and disable specific detection
  • UI: Ability to set language for Speech to Text and NLP processing
  • UI: AWS SageMaker Ground Truth integration allows you to create and manage your private work team and crowd source your labeling tasks
  • UI: Ability to set pagination to limit each query
  • UI: Parse JSON definition file and can ingest multiple files defined in the JSON file
  • UI: Player now shows markers of labels being detected
  • ES: Enhanced Elasticsearch engine, index all mediainfo, exif, and AIML metadata that allows you to do deep search
  • API: Simple APIs to ingest, analyze, and search content

Changed

  • replace Media Analysis solution analysis engine with new analysis engine

Removed

[1.0.0] - 2019-01-16

Added

  • initial checkin

Changed

Removed