Scenes-with-text Detection (v4.3)

About this version

  • Submitter: keighrim
  • Submission Time: 2024-04-11T21:49:21+00:00
  • Prebuilt Container Image: ghcr.io/clamsproject/app-swt-detection:v4.3

  • Release Notes

    This version brings many bug fixes and new models

    • fixed missing NEG label score in the classification property in TimePoint annotations (#87), and app metadata is updated accordingly
    • fixed sampling rate disparity (#90)
    • fixed sinusoidal positional features were not actually used (#47), and newly trained models with the fix are included
    • miscellaneous code clean-up

About this app (See raw metadata.json)

Detects scenes with text, like slates, chyrons and credits.

Inputs

(Note: “*” as a property value means that the property is required but can be any value.)

Configurable Parameters

(Note: Multivalued means the parameter can have one or more values.)

Name Description Type Multivalued Default Choices
startAt Number of milliseconds into the video to start processing integer N 0  
stopAt Number of milliseconds into the video to stop processing integer N 10000000  
sampleRate Milliseconds between sampled frames integer N 1000  
minFrameScore Minimum score for a still frame to be included in a TimeFrame number N 0.01  
minTimeframeScore Minimum score for a TimeFrame number N 0.5  
minFrameCount Minimum number of sampled frames required for a TimeFrame integer N 2  
modelName model name to use for classification string N 20240126-180026.convnext_lg.kfold_000 20240126-180026.convnext_lg.kfold_000, 20240212-131937.convnext_tiny.kfold_000, 20240212-132306.convnext_lg.kfold_000
useStitcher Use the stitcher after classifying the TimePoints boolean N true false, true
pretty The JSON body of the HTTP response will be re-formatted with 2-space indentation boolean N false false, true

Outputs

(Note: “*” as a property value means that the property is required but can be any value.)

(Note: Not all output annotations are always generated.)