Multi Channel Image

[szandara]

I am trying to work with multi-channel images. In OpenCV this is supported by the generic cv::Mat type. Where you can pass an arbitrary number of channels. The use case is to upscale an image after ML inference which has 7 channels. (I am using tract_onnx which also does not offer any resize operation on their Tensors).

I have been browsing the docs and examples but I cannot seem to figure out how to do that. From what I have seen, the Pixel data structure only holds maximum of 4 channels and supports just some well known image structures.

Is there a way to do it and if so how?

Thanks a lot for the help

[theotherphil]

Not sure how useful a reply is after three years (!), but…

This is not currently possible. I have no plans to move this crate off of the types defined in the image crate in the near future so this depends on making their Pixel types more generic. There’s an issue for this image-rs/image#1523 but I don’t expect much progress any time soon.