Instead of trying to work with the JPEG image, if your data is formatted correctly, you could have each person as a single entry in the data table, with the lat/long coordinates and all of the attributes of that person (Name, phone, village, filename of photo, etc.) as columns in your table. This could be imported as a spreadsheet and converted to X-Y data, so it create vector data (point shapefile) from the get-go instead of having to work with raster images.
To attach your geotagged photos to your point shapefile that contains all the demographic information, use Geotagged Photos to Points, under the Data Management Toolbox, using your shapefile as the output layer.
Once you have a shapefile with linked photos, you can use Conversion Tools' To KML Toolset to convert the shapefile into .kml, which Google Earth can use. I haven't done this myself, so I'm unsure of how the conversion to kml keeps the photos linked or requires linking them once in Google Earth.