Eduardo

Live555 does not decode the H264 data.You have to modify DummySink in module.cpp, applying ffmpeg decoding to extract decoded frames from the raw H264 data.

Well, there is a workaround I've been using for a while, try to override the _mode_segmentation_maps property of your augmenter: ```python aug._mode_segmentation_maps = aug.mode.value ``` And in the case you...

https://github.com/aleju/imgaug/issues/788

Any updates on this? Can we expect latency improvements for our pruned models?

The same here, it's able to train but it fails to reinitialize train after evaluation. Trace: ``` INFO:tensorflow:Running local_init_op. INFO:tensorflow:Done running local_init_op. INFO:tensorflow:Init TPU system INFO:tensorflow:Initialized TPU in 7 seconds...

Nothing still... It is a must for image augmentation pipelines: ```python image = tfa.image.gaussian_filter2d( image, sigma=tf.random.uniform([2], 0, 5, dtype=tf.float32), ) ```

As a workaround, you can monkeypatch the DateFinder class. ```python import datefinder class DateFinder(datefinder.DateFinder): def parse_date_string(self, date_string, captures): try: return super().parse_date_string(date_string, captures) except TypeError: return datefinder.DateFinder = DateFinder ```

It is not fixed, the behaviour persists for version v1.16.0. I think the PR #186 is still necessary.

The problem are the grouped convolutions https://github.com/huggingface/transformers/blob/a9eee2ffecc874df7dd635b2c6abb246fdb318cc/src/transformers/models/segformer/modeling_tf_segformer.py#L242-L244 It gets exported without errors at all when using standard convolutions (`groups=1`). Related to https://github.com/onnx/tensorflow-onnx/issues/2099 Any idea? workaround?

sure @Borda, I am sorry for the late response. It might take me some time, it is not straightforward to fix the affected tests, I'll do my best