Is it possible to process multiple regions of interest in detect and semanticseg?

Hi,

I'm experimenting with both the detect function for a FasterRCNN object and the semanticseg function for Semantic Segmentation to analyze an image. Detection and segmentation work fine. But for my application, it would be nice to be able couple to output of the detect function to semanticseg function. A possible input argument of both functions is a ROI, specified as a 4D vector. I've generated artificial training data and a ROI 'truth array' containing N instances. Thus I have a N x 4 dimensional array in the format that both functions expect.

Currently i'm getting an error, because both functions expect one ROI. One 1 x 4 vector.

Question: Is there a way to extend the functionality in order to process multiple ROI's in one shot?

I'm aware that the creation of a for-loop could process the same image several times with alternating ROI's but this not an elegant solution.

Thanks in advanced!

[Siddharth Jawahar]

I understand you want to process multiple regions of interest in both the detect and the semanticseg functions. As a precursor to this can you tell me a little about the training data? Are you training it to detect multiple objects in the image or just one? And also what is the exact error you are seeing when you try to pass in multiple rois and which release of MATLAB are you using?