Tao Gong

Hi~@noahcao, just a kind reminder in case you forgot it :)

Very appreciate your contribution. After discussing a general solution in your posted pr of mmdet, we can use the same behavior to fix the bug.

We have a plan to support it before the end of the year. By the way, we have already supported a multi-classes multi-object tracking dataset: YouTube-VIS dataset.

SELSA and SELSA with temporal ROI align need proposals generated by two stage detector. If you want to use yolox in vid, I suggest you use DFF or FGFA method

We will take it into our plan.

Hi~@khurramHashmi , very appreciate your contribution to support TF-Blender method. Please refer to [contributing guides](https://github.com/open-mmlab/mmtracking#contributing) in order to fix the lint error. Besides, we also have [unittests](https://github.com/open-mmlab/mmtracking/tree/master/tests) to verify the...

> Hello! We are the authors of this [paper](https://arxiv.org/pdf/2211.09108.pdf). We implemented our method fully based on MMTracking 0.x and experimented some more datasets like [UVO v0.5](https://sites.google.com/view/unidentified-video-object/dataset) and [OVIS](http://songbai.site/ovis/). I see...

The error reminds you that the `results` is a list rather than a dict. It means you send a list to the data pipeline. In the testing of mot, we...

In `regress_tracks()` function, bboxes of the previous frame will be propagated to the current frame. Please check whether the first dimension of [`bboxes`](https://github.com/open-mmlab/mmtracking/blob/master/mmtrack/models/mot/trackers/tracktor_tracker.py#L65) and [`ids`](https://github.com/open-mmlab/mmtracking/blob/master/mmtrack/models/mot/trackers/tracktor_tracker.py#L66) is the same. I guess...

Then, there is something wrong with the [`multiclass_nms()`](https://github.com/open-mmlab/mmtracking/blob/master/mmtrack/models/mot/trackers/tracktor_tracker.py#L72). You can check inside the function.