Hi Michél,
short answer first: as far as I can tell from the docs, MVI itself does not read or expose EXIF data. The models only care about pixels, and the metadata MVI stores with an image is its own (inference results, inspection outcomes, labels). I could not find anything in the MVI documentation or REST API that parses GPS tags automatically. The good news is twofold though: the location data usually survives inside the file, and MVI gives you a documented place to store it once you have extracted it.
The mechanism is the "user metadata" on dataset files. The MVI REST API lets you attach arbitrary key/value pairs to any file in a dataset, and you can then filter and query files by those keys (user_metadata.your-field-name in the files list API). The API reference documents it here:
https://public.dhe.ibm.com/systems/power/docs/powerai/api850.html
And IBM's official vision-tools CLI / Python library has dedicated resources for exactly this, fkeys and fmetadata:
https://github.com/IBM/vision-tools
So the pattern I would build:
1. Extract the GPS from the file before or alongside the MVI upload. For photos, exiftool is the gold standard (exiftool.org): something like exiftool -n -gpslatitude -gpslongitude -csv *.jpg gives you decimal coordinates ready to use, and it handles iPhone HEIC too. For iPhone videos the location is stored as a QuickTime tag rather than classic EXIF (an ISO 6709 location string), and exiftool reads that as well. If you prefer to keep it in code, Python with Pillow or exifread does the same job inside whatever script already talks to the MVI API.
2. Attach lat/long as user metadata to the file in MVI (via the API or vision-tools), so the dataset itself becomes searchable by location.
3. Push the coordinates to Maximo through the standard REST APIs. The typical landing spot, depending on your flow, is the service address latitude/longitude on the service request or work order you create from the inspection result, or the asset/location record itself. If you run Maximo Spatial, even better, that is exactly the data it wants.
Two caveats worth knowing. First, whether the GPS is still in the file depends on the path it took: photos straight from the iPhone camera roll keep it (if Location Services were on for the Camera app), but anything that went through WhatsApp, mail in compressed mode, or social media exports is usually stripped. Second, I have not found it documented whether every MVI upload path preserves the original file bit for bit, so before building the pipeline I would run a 5 minute test: upload one geotagged photo to a dataset, download it back with the dataset export, and run exiftool on it. If the tags survive, you can even extract them after the fact from files already sitting in MVI.
One last thought: if the images come from MVI Edge with fixed cameras, you usually already know the location per camera or inspection, and it is easier to inject it as a constant. The EXIF game really matters for mobile-captured content, which sounds like your iPhone case.
Regards
Roberto