I was wondering if it’s possible to add two input images to a workflow and have separate processes run on each simultaneously?
Basically, the images I am analysing come in sets of two. The processes which are run on image 2 are dependend on the output of the processes run on image 1.
For example: if image 1 is classified as “X” then detect this object class in image 2.
Unfortunately I see no way of adding two input images. The obvious work around is to combine the images and run object detection to separate them in the workflow, but that involves steps I’d rather avoid.
This is a great question, and it is totally possible within a workflow. You can configure the input block to take in 2 images!
Have you tried the Workflows AI Assistant before? It is very powerful and is capable of helping build your workflow. For example, I assembled a starter workflow using your verbiage in five minutes.
Thanks for your help. I have added second input image but unfortunately this leads to another issue. I wish to run quite a large number of processes on each image depending on whether they are classified as “A or B”. I have currently set up my workflow to run the processes I want on one image, but the only way I can see to run them all on the other image is to duplicate the entire chain of blocks for each image. I can’t see a way to link one model block to both image 1 or 2.
Similarly, I don’t seem to be able to base the “continue if” blocks on the outputs of multiple model blocks, like "[continue if] ([A/B classification of image 1 = A] OR [A/B classification of image 2] = A]
As there also doesn’t seem to be a way to encapsulate individual parts of the workflow, this is leading to my workflow becoming very convoluted.
I may end up having to run my individual model APIs in python just to keep everything legible.
Also, the AI assistant hasn’t been much help. It only gives one response before freezing and I can’t talk to it again until I exit and re-enter my workflow. If I ask it to make a change to the workflow, it just tells me how it’s going to do it, then freezes. I’m sure it will be a very powerful tool in the future, but this beta version currently isn’t functioning for me.