Update sequence model with new trained checkpoint (Seq_reweight5/checkpoint-best.pth)
Browse files- README.md +3 -53
- config.json +3 -4
- model.safetensors +1 -1
README.md
CHANGED
|
@@ -6,56 +6,6 @@ tags:
|
|
| 6 |
- pytorch_model_hub_mixin
|
| 7 |
---
|
| 8 |
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
```bibtex
|
| 13 |
-
@inproceedings{st4rtrack2025,
|
| 14 |
-
title={St4RTrack: Simultaneous 4D Reconstruction and Tracking in the World},
|
| 15 |
-
author={Feng*, Haiwen and Zhang*, Junyi and Wang, Qianqian and Ye, Yufei and Yu, Pengcheng and Black, Michael J. and Darrell, Trevor and Kanazawa, Angjoo},
|
| 16 |
-
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
|
| 17 |
-
year={2025}
|
| 18 |
-
}
|
| 19 |
-
```
|
| 20 |
-
|
| 21 |
-
## Model info
|
| 22 |
-
|
| 23 |
-
Github page: https://github.com/HavenFeng/St4rTrack-Release
|
| 24 |
-
Project page: https://st4rtrack.github.io/
|
| 25 |
-
Paper: https://arxiv.org/abs/2504.13152
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
## How to Use
|
| 29 |
-
|
| 30 |
-
First install St4rTrack. To load the model (by default checkpoint trained with Sequence Mode):
|
| 31 |
-
|
| 32 |
-
```python
|
| 33 |
-
from dust3r.model import AsymmetricCroCo3DStereo
|
| 34 |
-
|
| 35 |
-
# Loads default Seq checkpoint
|
| 36 |
-
model = AsymmetricCroCo3DStereo.from_pretrained("yupengchengg147/St4rTrack")
|
| 37 |
-
|
| 38 |
-
```
|
| 39 |
-
|
| 40 |
-
run following code to load checkpoint trained with Pair Mode:
|
| 41 |
-
```python
|
| 42 |
-
from huggingface_hub import hf_hub_download
|
| 43 |
-
import tempfile
|
| 44 |
-
import os
|
| 45 |
-
from dust3r.model import AsymmetricCroCo3DStereo
|
| 46 |
-
|
| 47 |
-
# Create a temporary directory for the model files
|
| 48 |
-
temp_dir = os.path.join(tempfile.gettempdir(), "st4rtrack_pair")
|
| 49 |
-
os.makedirs(temp_dir, exist_ok=True)
|
| 50 |
-
|
| 51 |
-
# Download the config and model files from the Pair subfolder
|
| 52 |
-
config_path = hf_hub_download(
|
| 53 |
-
repo_id="yupengchengg147/St4rTrack",
|
| 54 |
-
filename="Pair/config.json",
|
| 55 |
-
cache_dir=temp_dir
|
| 56 |
-
)
|
| 57 |
-
# Load the model from the downloaded path
|
| 58 |
-
model_dir = os.path.dirname(config_path)
|
| 59 |
-
model = AsymmetricCroCo3DStereo.from_pretrained(model_dir)
|
| 60 |
-
|
| 61 |
-
```
|
|
|
|
| 6 |
- pytorch_model_hub_mixin
|
| 7 |
---
|
| 8 |
|
| 9 |
+
This model has been pushed to the Hub using the [PytorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration:
|
| 10 |
+
- Library: https://github.com/naver/dust3r
|
| 11 |
+
- Docs: [More Information Needed]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
config.json
CHANGED
|
@@ -24,12 +24,11 @@
|
|
| 24 |
512,
|
| 25 |
288
|
| 26 |
],
|
| 27 |
-
"landscape_only":
|
| 28 |
"output_mode": "pts3d",
|
| 29 |
-
"patch_embed_cls": "
|
| 30 |
"pos_embed": "RoPE100",
|
| 31 |
"random_init": "none",
|
| 32 |
"random_init_lr_scale": 1.0,
|
| 33 |
-
"rope_mode": "full_3d"
|
| 34 |
-
"seq_mode": "NA"
|
| 35 |
}
|
|
|
|
| 24 |
512,
|
| 25 |
288
|
| 26 |
],
|
| 27 |
+
"landscape_only": true,
|
| 28 |
"output_mode": "pts3d",
|
| 29 |
+
"patch_embed_cls": "ManyAR_PatchEmbed",
|
| 30 |
"pos_embed": "RoPE100",
|
| 31 |
"random_init": "none",
|
| 32 |
"random_init_lr_scale": 1.0,
|
| 33 |
+
"rope_mode": "full_3d"
|
|
|
|
| 34 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2284790056
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba6d8afc4b65789c12f30dceb66ef9c6076c3a72deeff22024380c7883f1a007
|
| 3 |
size 2284790056
|