File size: 663 Bytes
04b47c7
 
 
 
 
 
 
dd8f9f1
04b47c7
dd8f9f1
04b47c7
 
dd8f9f1
04b47c7
dd8f9f1
 
04b47c7
dd8f9f1
 
 
 
04b47c7
 
 
 
dd8f9f1
 
04b47c7
dd8f9f1
04b47c7
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
---
license: cc-by-nc-4.0
tags:
- video-object-segmentation
- video-segmentation
- evaluation
---

# FaVOS

FaVOS evaluation uses the first annotated frame as the prompt frame. Metrics are
computed on the remaining frames only.

Download the dataset:

```bash
huggingface-cli download FaVOSsubmission/FaVOS --repo-type dataset --local-dir data
unzip data/JPEGImages.zip -d data
unzip data/Annotations.zip -d data
```

Dataset files are available at:
https://huggingface.co/datasets/FaVOSsubmission/FaVOS/tree/main

Evaluate predictions against:

```text
data/Annotations/<video_id>/<frame>.png
```

Use `favos-20.txt` and `favos-40.txt` as the evaluation splits.