Nikita Karaev

Hi @pvtoan, this will take some time, I'll try to take a look over the weekend!

Hi @mikebilly, you can check out our [Colab demo](https://colab.research.google.com/github/facebookresearch/co-tracker/blob/master/notebooks/demo.ipynb) that shows how to track a single query point from any frame in a video. Please look at the section "Tracking...

Hi @fangli333, thank you for the question! Backward tracking is only supported in the offline mode because you wouldn't need an online algorithm if you already have access to all...

Hi @Davidyao99, thank you! Could you open a PR?

Hi @dat-nguyenvn, the released model doesn't have a confidence flag. There's currently no way to filter out bad tracks because of that.

Hi @StephenZhao1, sorry for the late response. I think the model can handle such objects, but not always. Could you please send an example of such a video? This way,...

Hi @rogerioagjr, apologies for the delayed response. We've made some changes to CoTracker for the release and retrained the model on Kubric, but not on PointOdyssey. So, the code for...

Hi @qianduoduolr, these are the settings that we used to prepare 11000 training sequences for Kubric MOVI-f: ``` create_point_tracking_dataset( train_size=(512, 512), shuffle_buffer_size=None, split="train", batch_dims=tuple(), repeat=True, vflip=False, random_crop=True, tracks_to_sample=2048, sampling_stride=4, max_seg_id=25,...

As far as I remember, we had to download the dataset first for it to work. Then, we load MOVI-f (not MOVI-e) from a local path.

Hi @Anderstask1 and @NEbrahimi, I apologize for missing your messages as the issue was closed. I modified this function: https://github.com/google-research/kubric/blob/e140e24e078d5e641c4ac10bf25743059bd059ce/challenges/point_tracking/dataset.py#L992 Here are the changes I made: ``` import torch from...