IssueTracker

Please describe your requested enhancement. Good feature requests will solve common problems or enable new use cases.

What you would like to accomplish:
- Customer would like to use ML Engine Online Predictions on image data which exceeds the prediction request payload size [1](since image data easily exceeds 1.5 MB).

[1] https://cloud.google.com/ml-engine/docs/scikit/using-pipelines-for-preprocessing#send_online_prediction_request


How this might work:
- If the request limit size can be increased to higher limit or the product team can add better image processing support for Online Prediction, then it would be great.


If applicable, reasons why alternative solutions are not sufficient:
- a potential way to workaround this limit is by splitting up the data into multiple prediction requests, and having ML Engine wait and aggregate the data once all of the multiple prediction requests have been sent; unfortunately this is not currently available, and the prediction results will be processed and returned right when the request is sent.


Other information (workarounds you have tried, documentation consulted, etc):