For some not entirely well defined reasons, I originally copied images from the ingestion bucket to the animl-images-serving bucket (from which the frontend accesses them), but also to an animl-images-archive bucket, where they were a little easier to track down and access because the object keys included the camera serial numbers as a top level "directory". The current archive bucket key template looks like: <cameraId>/<originalFileName>_<imageId>.<ext>.
Now that we are allowing users to (a) override camera serial numbers and (b) not requiring that camera serial numbers be unique across the entire app (although they will need to be unique for a given project), we would have to update that template to include the projectId in there (something like <projectId>/<cameraId>/<originalFileName>_<imageId>.<ext> would probably work`).
Regardless, I think it's the archive is redundant and it's time to abandon it.
I think all that needs to be done in this code-base is to remove the copy_to_archive() function and calls, but this also has implications for the COCO export in animl-api and the download images script in animl-analytics.
For some not entirely well defined reasons, I originally copied images from the ingestion bucket to the
animl-images-servingbucket (from which the frontend accesses them), but also to ananiml-images-archivebucket, where they were a little easier to track down and access because the object keys included the camera serial numbers as a top level "directory". The current archive bucket key template looks like:<cameraId>/<originalFileName>_<imageId>.<ext>.Now that we are allowing users to (a) override camera serial numbers and (b) not requiring that camera serial numbers be unique across the entire app (although they will need to be unique for a given project), we would have to update that template to include the projectId in there (something like
<projectId>/<cameraId>/<originalFileName>_<imageId>.<ext>would probably work`).Regardless, I think it's the archive is redundant and it's time to abandon it.
I think all that needs to be done in this code-base is to remove the
copy_to_archive()function and calls, but this also has implications for the COCO export in animl-api and the download images script in animl-analytics.