#### NOTE
This is a **community plugin**, an external project maintained by its respective author.
Community plugins are not part of FiftyOne core and may change independently.
Please review each plugin’s documentation and license before use.
# mose-v2
[](https://discord.gg/fiftyone-community)
[](https://huggingface.co/Voxel51)
[](https://voxel51.com/blog)
[](https://share.hsforms.com/1zpJ60ggaQtOoVeBqIZdaaA2ykyk)
[](https://www.linkedin.com/company/voxel51)
[](https://x.com/voxel51)
[](https://medium.com/voxel51)
A [FiftyOne remote zoo dataset](https://docs.voxel51.com/dataset_zoo/remote.html) integration for **MOSEv2**, a large-scale video object segmentation benchmark: thousands of videos, instance masks, and diverse real-world conditions (occlusion, small objects, weather, low light, camouflage, etc.). See the [project site](https://mose.video/) and [upstream repo](https://github.com/FudanCVL/MOSEv2) for the full benchmark description.
## Source and citation
- **Website**: [mose.video](https://mose.video/)
- **GitHub**: [MOSEv2](https://github.com/FudanCVL/MOSEv2)
- **Hugging Face (dataset card)**: [FudanCVL/MOSEv2](https://huggingface.co/datasets/FudanCVL/MOSEv2)
- **License**: Original MOVEv2 terms: [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)
- **Citation**: See also other related citations from the [official MOSEv2 website](https://mose.video/#citation).
```default
@article{MOSEv2,
title={{MOSEv2}: A More Challenging Dataset for Video Object Segmentation in Complex Scenes},
author={Ding, Henghui and Ying, Kaining and Liu, Chang and He, Shuting and Jiang, Xudong and Jiang, Yu-Gang and Torr, Philip HS and Bai, Song},
journal={arXiv preprint arXiv:2508.05630},
year={2025}
}
```
## Quick start
Installation
```bash
pip install fiftyone
pip install gdown # required for Google Drive download; see also requirements.txt
```
Load via the FiftyOne Dataset Zoo
```python
import fiftyone as fo
import fiftyone.zoo as foz
dataset = foz.load_zoo_dataset(
"https://github.com/voxel51/mose-v2",
split="train", # or "validation"
max_samples=1000, # optional, for quicker exploration
)
session = fo.launch_app(dataset)
# For a dynamic Grouped view
grouped_view = dataset.group_by("sequence_id", order_by="frame_number")
```
## Notes:
- Downloads **train** and **validation** archives from Google Drive (file IDs are in `__init__.py` as `DRIVE_FILE_IDS`).
- Extracts `train/` and `valid/` under the FiftyOne-managed dataset directory. A symlink `validation` → `valid` is created when needed so split names match FiftyOne’s expectations.
```text
dataset_dir/
train/
JPEGImages//{00000,00001,...}.jpg
Annotations//{00000,00001,...}.png
valid/
JPEGImages//{00000,00001,...}.jpg
Annotations//00000.png
```
- Registers **one sample per video frame**. Segmentation is stored as an indexed PNG per frame (`ground_truth`: `fo.Segmentation` with `mask_path`).
- Annotation masks are **8-bit indexed PNGs**: pixel value `0` is background; value `N` is object instance `N`.
## Sample fields
| Field | Role |
|----------------|----------------------------------------------------|
| `filepath` | Path to the JPEG frame |
| `sequence_id` | Video sequence name |
| `frame_number` | Zero-based frame index |
| `tags` | Split and sequence (e.g. `train`, sequence id) |
| `ground_truth` | `Segmentation` with `mask_path` to the indexed PNG |
## Statistics
| Split | Sequences | Total Samples | Annotated Samples |
|------------|-------------|-----------------|------------------------|
| train | 3,666 | 311,843 | 311,843 |
| validation | 433 | 66,526 | 433 (first frame only) |
## Visualize
Each image is tagged with its **split** and with its **sequence** name — frames that share a `sequence_id` belong to the same clip.
For a **video-like** browser in the App, use a dynamic grouped view — one group per sequence, frames ordered by `frame_number`.
