Christoph Heindl

Hey! True, didn't notice this difference between the methods. The 5x5 is the only method that clamps values before returning https://github.com/cheind/pytorch-debayer/blob/master/debayer/modules.py#L333 I added this AFAIK as the kernels weren't scale...

hey, so you got a line-based camera? The arrangement of filters is currently unsupported by this library out of the box. You will need to tweak things a bit. You...

@follesoe are you saying that you can sucessfully restart the preview stream after triggering the shutter? I'm testing this on a Hero12 using the WiredGoPro (USB) interface and it seems...

> @follesoe are you saying that you can sucessfully restart the preview stream after triggering the shutter? I'm testing this on a Hero12 using the WiredGoPro (USB) interface and it...

thanks!

Hey @hongin13, does this statement in Readme.md resolve your question? > Metric MOTP seems to be off. To convert compute (1 - MOTP) * 100. [MOTChallenge](https://motchallenge.net/) benchmarks compute MOTP as...

@Justin900429 thanks a lot! py-motmetrics can cache intermediate results. If you look here https://github.com/cheind/py-motmetrics/blob/develop/motmetrics/metrics.py#L204 a cache is being setup for all intermediate metrics that are computed, which is then being...

@Justin900429 that is a good question. I haven't looked at HOTA yet, but if it the computation is similar to Average Precision (AP) metric, one could instead pre-compute a 'tensor'...

@Justin900429 thanks for your update! Would it be feasible to implement HOTA like AP is implemented for COCO datasets? See here https://github.com/ppwwyyxx/cocoapi/blob/4670067b35e7b65d618c9746c456fe2c1128049c/PythonAPI/pycocotools/cocoeval.py#L315 where in `accumulate` a tensor with multiple axis...

@Justin900429 amazing work - thanks a lot for your time and effort! Just merged into develop :)