100 uploads to sharepoint: Triggered flow does not run through all changes --> trigger, apply to each questions

[Teflonreis]

Hiiii

Short:

I have build flows so far that get triggered by 1! change and run through 1!/the same change.

The latest flow struggles with bulk changes. My trigger or apply to each must be wrong.

Case:

I created a sharepoint list where I upload 100 images. Each image shall be labeled according to 10 AI builder models.

e.g. "Is the image an outdoor image or indoor image" is one categorizing AI builder model.

Second AI builder model is "image contains cables" or "image contains no cables"

...

There are 10 models. Each uploaded image shall run though all 10 models using power automate.

Without apply to each, only a couple images were successfully labelled but many images and models left out. Only the first model "Detect indoor or outdoor images and second one "bushing" worked a bit.

In close up

Tried using an additional action "Get changes for an item or a file(properties only) and an apply to each where all models are contained. Got errors. Maybe chose the wrong output from previous steps in apply to each.

A hint is greatly appreaciated.

Many thanks for reading my post. 

[okeks]

I'm not sure if the following also applied to the Sharepoint connector because there is nothing about this in the documentation, but as Onedrive for Business is based on Sharepoint I can imagine that this limitations also hit the Sharepoint connector:

11. On new file and on modified file triggers (all variants) currently may have issues when there are more than approximately 30 pending changes between two trigger polls.

Source: OneDrive for Business - Connectors | Microsoft Docs

[Teflonreis]

Oh thanks for that info. 
I also noticed that my flow recognises the image categories, but does not add a tag in sharepoint.

Appperently, my images in sharepoint were tagged my microsoft themselves.  Sharepoint image tagging 

I want to build a flow that does the tagging like microsoft, but using my Ai builder models.