OpenCV stitching seems to not work

[cocoa-xu]

OpenCV stitching seems to not work

[cocoa-xu]

Hi, thank you for trying this library! (Also, I checked the stream, that was a great one!)

I can confirm this was a bug when handling one or more enum type arguments that appeared in the function call, for example.

CV_WRAP static Ptr<Stitcher> create(Mode mode = Stitcher::PANORAMA);

d28b1d5 should fix this issue (and other similar cases) by

• declaring their types as their underlying types in the generated code. std::underlying_type_t<Mode> mode = ...
• static casting them back to the original type when using them. static_cast<Mode>(...)

[lawik]

Sweet.

Use whatever you like from the stuff I did in the stream :)

Very curious to see if the stitching stuff works well.

Want to find a good example of improving photos as well. That camera is pretty crappy.

I don't know if haar cascades are still a relevant way for finding faces but I'd want some kind of face tracking. That I can actually do extra neat things with. ML isn't my space so now idea where the best bang for the buck is today.

[lawik]

Sounds great!

I was amazed at what worked. Hope the stream did your work some level of justice.

Let me know if you have some examples you think I should try. I have a decent chunk of Nerves devices and media gear.

[cocoa-xu]

Haha, to be honest, I was surprised that a number of functions just worked as we would expect!

I wonder if it is possible for me to use some of your code as examples here? Or, if you would like to create a PR please feel free to do so!

One more thing, in my test run, I observed that the return value was something like {:ok, {:ok, 0}, #Reference} , where the 0 was one of the values in the Status enum in C++.

a099f2a fixed the return value to what it was supposed to be (the correct result should be something like {:ok, 0, #Reference}).

[cocoa-xu]

The API works as expected now, and here is an example. I'm gonna close this issue, but feel free to reopen it if there are any glitches. :)