-
Notifications
You must be signed in to change notification settings - Fork 50
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
InceptionResnetV2 feature for Query-by-Example #243
Conversation
…s rather than only the most representative frame.
Visual-text co-embedding video extraction
This is required since FFMpegAudioDecoder and TechnicalVideoMetadataExtractor use Math.floorDiv, which is a function only available since Java 9.
Due to popular demand.
Updated log4j *again* to 2.16
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good to me
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I generally agree that it makes much more sense to have the output of the resnet as a feature directly. I only have one minor comment regarding the method used for image scaling, see inline comment.
cineast-core/src/main/java/org/vitrivr/cineast/core/features/InceptionResnetV2.java
Outdated
Show resolved
Hide resolved
…s rather than only the most representative frame.
This is required since FFMpegAudioDecoder and TechnicalVideoMetadataExtractor use Math.floorDiv, which is a function only available since Java 9.
Due to popular demand.
# Conflicts: # cineast-core/src/main/java/org/vitrivr/cineast/core/features/VisualTextCoEmbedding.java # gradle.properties
# Conflicts: # build.gradle # cineast-core/src/main/java/org/vitrivr/cineast/core/features/VisualTextCoEmbedding.java
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I bumped the version number to v3.6.2, since that's what the next minor release would be
cineast-core/src/main/java/org/vitrivr/cineast/core/features/InceptionResnetV2.java
Show resolved
Hide resolved
…nstead of only using most representative frame.
cineast-core/src/main/java/org/vitrivr/cineast/core/features/InceptionResnetV2.java
Outdated
Show resolved
Hide resolved
…tV2 and VisualTextCoEmbedding.
Refactored InceptionResnetV2 feature encoding into its own separate feature from inside VisualTextCoEmbedding.
This allows it to be used as a feature for query-by-example that is less abstracted towards semantic content than the visual-text co-embedding.