---
library_name: pytorch
tags:
  - video-classification
  - vision-transformer
  - trackmae
  - videomae
---

# TrackMAE Checkpoints

Pretrained and fine-tuned video checkpoints from
[TrackMAE: Video Representation Learning via Track Mask and Predict](https://arxiv.org/abs/2603.27268).
The weights are PyTorch state dictionaries compatible with the official
[TrackMAE codebase](https://github.com/rvandeghen/TrackMAE).

## Pretrained Checkpoints

| File | Backbone | Dataset | Epochs | Spatial Target |
| --- | --- | --- | --- | --- |
| `pretrain/vit_b/videomae_cotracker_base_bs_512_mask_tube_800_lambda_0.25_up_14_clip.pth` | ViT-B | Kinetics-400 | 800 | CLIP |
| `pretrain/vit_l/videomae_cotracker_large_bs_1024_mask_tube_800_lambda_0.25_up_14_clip_k700.pth` | ViT-L | Kinetics-700 | 800 | CLIP |


## Fine-Tuned Checkpoints

| File | Backbone | Pretraining | Fine-tuning |
| --- | --- | --- | --- |
| [`finetune/vit_b/k400_finetuned_vit_b_with_k400_pretraining_trackmae.pth`](https://huggingface.co/rvandeghen/TrackMAE/blob/main/finetune/vit_b/k400_finetuned_vit_b_with_k400_pretraining_trackmae.pth) | ViT-B | Kinetics-400 | Kinetics-400 |
| [`finetune/vit_b/k400_finetuned_vit_b_with_k400_pretraining_trackmae_with_spatial_pixel_targets.pth`](https://huggingface.co/rvandeghen/TrackMAE/blob/main/finetune/vit_b/k400_finetuned_vit_b_with_k400_pretraining_trackmae_with_spatial_pixel_targets.pth) | ViT-B | Kinetics-400 | Kinetics-400 |
| [`finetune/vit_b/ssv2_finetuned_vit_b_with_k400_pretraining_trackmae.pth`](https://huggingface.co/rvandeghen/TrackMAE/blob/main/finetune/vit_b/ssv2_finetuned_vit_b_with_k400_pretraining_trackmae.pth) | ViT-B | Kinetics-400 | Something-Something V2 |
| [`finetune/vit_b/ssv2_finetuned_vit_b_with_k400_pretraining_trackmae_with_spatial_pixel_targets.pth`](https://huggingface.co/rvandeghen/TrackMAE/blob/main/finetune/vit_b/ssv2_finetuned_vit_b_with_k400_pretraining_trackmae_with_spatial_pixel_targets.pth) | ViT-B | Kinetics-400 | Something-Something V2 |
| [`finetune/vit_b/ssv2_finetuned_vit_b_with_ssv2_pretraining_trackmae.pth`](https://huggingface.co/rvandeghen/TrackMAE/blob/main/finetune/vit_b/ssv2_finetuned_vit_b_with_ssv2_pretraining_trackmae.pth) | ViT-B | Something-Something V2 | Something-Something V2 |
| [`finetune/vit_l/ssv2_finetuned_vit_l_with_k700_pretraining_trackmae.pth`](https://huggingface.co/rvandeghen/TrackMAE/blob/main/finetune/vit_l/ssv2_finetuned_vit_l_with_k700_pretraining_trackmae.pth) | ViT-L | Kinetics-700 | Something-Something V2 |
| [`finetune/vit_l/k400_finetuned_vit_l_with_k700_pretraining_trackmae.pth`](https://huggingface.co/rvandeghen/TrackMAE/blob/main/finetune/vit_l/k400_finetuned_vit_l_with_k700_pretraining_trackmae.pth) | ViT-L | Kinetics-700 | Kinetics-400 |

## Usage

Download a pretrained checkpoint:

```python
from huggingface_hub import hf_hub_download

checkpoint = hf_hub_download(
    repo_id="rvandeghen/TrackMAE",
    filename="pretrain/vit_b/videomae_cotracker_base_bs_512_mask_tube_800_lambda_0.25_up_14_clip.pth",
)
```


## Citation

```bibtex
@inproceedings{vandeghen2026trackmae,
  title     = {TrackMAE: Video Representation Learning via Track Mask and Predict},
  author    = {Vandeghen, Renaud and Thoker, Fida Mohammad and Van Droogenbroeck, Marc and Ghanem, Bernard},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {2026}
}
```