search

IBM / watson-multimedia-analyzer

WARNING: This repository is no longer maintained ⚠️ This repository will not be updated. The repository will be kept available in read-only mode. A Node app that use Watson Visual Recognition, Speech to Text, Natural Language Understanding, and Tone Analyzer to enrich media files.
https://developer.ibm.com/code/patterns/enrich-multi-media-files-using-ibm-watson/
Apache License 2.0
23 stars 25 forks source link

Bower packages breaks the rightside moment-viewer #39

Open haviduck opened 6 years ago

haviduck commented 6 years ago

Doing a straight bower install, the bower packages will break the moment viewer. Launch an instance, start the gui and click "Show moment" will throw this error:

angular.js:15018 TypeError: Failed to set the 'currentTime' property on 'HTMLMediaElement': The provided double value is non-finite. at Object.seekTime (videogular.js:405) at VideoController.js:66 at Scope.$broadcast (angular.js:19165) at MomentController.vm.gotoMoment (watson.moment.directive.js:68)

This is confirmed on win 10 and ubuntu 18.04 + 16.04.

haviduck commented 6 years ago

Turns out its a well known issue after moving over to AMI api keys. visualrecognition.js v3 doesn't look like it copes with it.

rhagarty commented 6 years ago

@haviduck - this code pattern has known issues when using new instances of Visual Recognition that require IAM api keys. There are also some API changes in v3 that break the way we process images.

Hopefully, someone will be able to address these issues shortly.

haviduck commented 6 years ago

@rhagarty - yeah, took me some deep diving to figure it all out. i'm very close to a fix - i just gotta find whatever is STILL triggering the old faces model. its being logged and im finding it hard to figure out how to update the Faces GA. when done ill commit and u can check it out

haviduck commented 6 years ago

nah. its not all about the faces.. that is sorted - but they removed the endtime's on speechtotext. if you use speakers that'll give the endtime per scene, so working on that.