Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,35 @@
|
|
1 |
-
---
|
2 |
-
license: mit
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
task_categories:
|
4 |
+
- object-detection
|
5 |
+
size_categories:
|
6 |
+
- 1K<n<10K
|
7 |
+
---
|
8 |
+
|
9 |
+
# TF-ID arXiv papers dataset
|
10 |
+
- This is the dataset for finetuning [TF-ID](https://huggingface.co/yifeihu/TF-ID-base) models.
|
11 |
+
- It contains about 4,600 images (academic paper pages) with bounding boxes of tables and figures in **coco format**.
|
12 |
+
- The papers are selected from [Hugging Face Daily Papers](https://huggingface.co/papers), covering mostly AI/ML/DL related topics.
|
13 |
+
- You can use this dataset to reproduce all TF-ID models.
|
14 |
+
- All bounding boxes were annotated manually by [Yifei Hu](https://x.com/hu_yifei)
|
15 |
+
|
16 |
+
## Project Repo
|
17 |
+
[github.com/ai8hyf/TF-ID](https://github.com/ai8hyf/TF-ID)
|
18 |
+
|
19 |
+
## Variants
|
20 |
+
Unzip the arxiv_paper_images.zip to get all images for training and testing.
|
21 |
+
The example below shows the difference between `annotations_with_caption.json` and `annotations_no_caption.json`:
|
22 |
+
![image/png](https://huggingface.co/yifeihu/TF-ID-base/resolve/main/td-id-caption.png)
|
23 |
+
|
24 |
+
## Citation
|
25 |
+
If you find TD-ID useful, please cite this project as:
|
26 |
+
```
|
27 |
+
@misc{TF-ID,
|
28 |
+
author = {Yifei Hu},
|
29 |
+
title = {TF-ID: Table/Figure IDentifier for academic papers},
|
30 |
+
year = {2024},
|
31 |
+
publisher = {GitHub},
|
32 |
+
journal = {GitHub repository},
|
33 |
+
howpublished = {\url{https://github.com/ai8hyf/TF-ID}},
|
34 |
+
}
|
35 |
+
```
|