flutter_tflite VS dio

For imageMean and imageStd, looking at the android implementation, we can see that these are passed along from dart to java eventually used at this point. Looking at the rest of this method, we can see this is in the context of converting the image data pixel by pixel into a bytearray that is eventually fed into the android tensorflow library. The actual conversion of pixel to bytearray is done by looking at the pixel RGB values separately and subtracting the mean then dividing by the std. This is meant to normalize your input data into "reasonable" data for your model.

With no release in the last 6 months, piling up of issues and launch of a new fork (diox), I felt something was going on with dio. Sadly, today I came across an official announcement (updated on the project repo 2 days back):

You might want to think twice about dio....we evaluated dio on a recent project and decided to avoid it after running into a blocker when setting a custom timeout (this bug: https://github.com/flutterchina/dio/issues/1454)

- and Dio for http requests

I also had some troubles choosing which Dart HTTP client to use. I've started with dio, but when I tried to read one of our bigger JSON responses (~1.5mb) my browser tab froze (there seems to be an old issue describing the same behavior). After that I've switched to the default dart's http package and it is working fine but it doesn't have builtin requests abort/cancellation.