Pretrained checkpoints and custom video example

[v-iashin]

Hi.

First of all, the repository provides clear and working instructions on how to pre-process the data, and how to do training and evaluation. Thanks!

While I could initialize training using your instructions, I was wondering if you have plans to release the pre-trained checkpoints for all three tasks. It would be nice not to spend compute on reproducing these.

In addition, it would be great if you could provide an example of how to use the pre-trained checkpoints to analyse a .mp4 file

Anyway, great work and thank you for making it open-source.

[ShirleyMaxx]

Hi,

Thank you for your kind words and for taking the time to explore our repository. We’re glad you found the instructions clear and helpful!

1. For pre-trained checkpoints, we’re sorry that we are unable to upload all the checkpoints for every task due to the large number of models. However, if you’ve prepared the dataset and set up the environment as provided in our instructions, training should be very efficient—it typically takes only 1-2 days to train each model. All the training commands are already included in the RUN_*.md file under the scripts directory, so you should have everything you need to get started.

2. As for analyzing .mp4 files, we didn’t provide a direct tool for this in the current repository. However, if you’re interested, we encourage you to try AlphaChimp, our newly proposed model that significantly outperforms the current baselines. AlphaChimp supports inference on .mp4 videos and can directly generate tracking and action recognition results.

Let us know if you have any further questions or need additional guidance. Thanks again for your interest!

[ShirleyMaxx]

Hi, @v-iashin here are the pretrained weights of most models for the three tracks. Please download them if needed.