[
https://issues.apache.org/jira/browse/BEAM-9247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17036100#comment-17036100
]
Kamil Wasilewski commented on BEAM-9247:
----------------------------------------
I think in some cases it makes sense to accept both image and image_context in
the form of tuple as an input. But, on the other hand, only few Video
Intelligence and Cloud Vision features require additional parameters. In this
case, video_context/image_context is always empty.
What about creating two public PTransforms? The first transform could let the
user specify one unique context per element and the second transform could
accept the simpler form of an input (without context). Then, the user could
choose between better control over the input data or simplicity.
In terms of implementation, these transforms could use the same DoFn
underneath.
Let me know what you think about this.
> [Python] PTransform that integrates Cloud Vision functionality
> --------------------------------------------------------------
>
> Key: BEAM-9247
> URL: https://issues.apache.org/jira/browse/BEAM-9247
> Project: Beam
> Issue Type: Sub-task
> Components: io-py-gcp
> Reporter: Kamil Wasilewski
> Assignee: Elias Djurfeldt
> Priority: Major
>
> The goal is to create a PTransform that integrates Google Cloud Vision API
> functionality [1].
> [1] https://cloud.google.com/vision/docs/
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
