This mobile AI benchmark are create offer multiple testing methodology, such as
* Standard Benchmark(Classification, Detection, Segmentation, Speech Recognition, Super Resolution)
* Out-Of-Order Benchmark(Prevent vendors use pattern matching to cheat on testing result)
* Pressure Test( Evaluate the stability of the AI performance )
etc...
We are still developing new test methodology to make benchmark result are as close as the real world performance.

If you have any benchmark idea or anything about this platform want to discuss with us, feel free to contact us by email.