Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pretty_name: SPEC
|
| 3 |
+
task_categories:
|
| 4 |
+
- image-to-text
|
| 5 |
+
- text-to-image
|
| 6 |
+
- image-classification
|
| 7 |
+
tags:
|
| 8 |
+
- image
|
| 9 |
+
- text
|
| 10 |
+
language:
|
| 11 |
+
- en
|
| 12 |
+
license: apache-2.0
|
| 13 |
+
size_categories:
|
| 14 |
+
- 1K<n<10K
|
| 15 |
+
---
|
| 16 |
+
|
| 17 |
+
# [CVPR 2024] SPEC Benchmark: Evaluating VLMs in Fine-grained and Compositional Understanding
|
| 18 |
+
introduced in the CVPR 2024 paper [Synthesize, Diagnose, and Optimize: Towards Fine-Grained Vision-Language Understanding](https://huggingface.co/papers/2312.00081)
|
| 19 |
+
|
| 20 |
+
[**Code**](https://github.com/wjpoom/SPEC) | [**🤗 Paper**](https://huggingface.co/papers/2312.00081) | [**📖 arXiv**](https://arxiv.org/abs/2312.00081)
|
| 21 |
+
|
| 22 |
+
To evaluate the understanding capability of visual-language models on fine-grained concepts, we propose a new benchmark, SPEC,
|
| 23 |
+
which consists of six distinct subsets, distributed across the dimensions of **S**ize, **P**osition, **E**xistence, and **C**ount.
|
| 24 |
+
Each test case consists of an image candidate set, which differs only in certain visual concepts, and a text candidate set,
|
| 25 |
+
which differs only in the corresponding language concept.
|
| 26 |
+
<p align="center">
|
| 27 |
+
<img src="https://cdn-uploads.huggingface.co/production/uploads/649bce4f200e2dff194d9883/sE65-zVjY_HXUT4-eaqZ9.png" width="90%"/>
|
| 28 |
+
<be>
|
| 29 |
+
</p>
|
| 30 |
+
|
| 31 |
+
## 🔧 Usage
|
| 32 |
+
### install
|
| 33 |
+
``` shell
|
| 34 |
+
git clone https://github.com/wjpoom/SPEC.git
|
| 35 |
+
cd SPEC/
|
| 36 |
+
pip install -e .
|
| 37 |
+
```
|
| 38 |
+
### prepare data
|
| 39 |
+
* run the following code in Python shell, replace `/path/to/save/data` with a specified dir to store the data.
|
| 40 |
+
```python
|
| 41 |
+
import zipfile
|
| 42 |
+
import os
|
| 43 |
+
from huggingface_hub import hf_hub_download
|
| 44 |
+
|
| 45 |
+
data_root = '/path/to/save/data'
|
| 46 |
+
hf_hub_download(repo_id='wjpoom/SPEC', repo_type='dataset', filename='data.zip', local_dir=data_root)
|
| 47 |
+
|
| 48 |
+
with zipfile.ZipFile(os.path.join(data_root, 'data.zip'), 'r') as zip_ref:
|
| 49 |
+
zip_ref.extractall(os.path.join(data_root))
|
| 50 |
+
|
| 51 |
+
os.remove(os.path.join(data_root, 'data.zip'))
|
| 52 |
+
```
|
| 53 |
+
### explore the dataset
|
| 54 |
+
* We provide a 📓notebook that enables you to visually explore the test samples in the SPEC dataset.
|
| 55 |
+
* Run this notebook either [locally](https://github.com/wjpoom/SPEC/blob/main/notebooks/explore_spec_local.ipynb) or online using [Colab](https://colab.research.google.com/github/wjpoom/SPEC/blob/main/notebooks/explore_spec_colab.ipynb).
|
| 56 |
+
|
| 57 |
+
### reproduce the results
|
| 58 |
+
* In our paper, we evaluated four popular VLMs using our SPEC dataset, namely: CLIP, BLIP, FLAVA and CoCa.
|
| 59 |
+
* To reproduce the results with these VLMs, you can run [this script](https://github.com/wjpoom/SPEC/blob/main/spec/run_eval.sh).
|
| 60 |
+
* You can also reproduce with this [local notebook](https://github.com/wjpoom/SPEC/blob/main/notebooks/evaluate_example_local.ipynb) or the online [Colab notebook](https://colab.research.google.com/github/wjpoom/SPEC/blob/main/notebooks/evaluate_example_colab.ipynb).
|
| 61 |
+
|
| 62 |
+
### evaluate custom VLMs
|
| 63 |
+
* If you want to evaluate your custom model on SPEC, you can follow the instructions in [this document](https://github.com/wjpoom/SPEC/blob/main/docs/evaluate_custom_model.md).
|
| 64 |
+
|
| 65 |
+
* ## ✒️ Citation
|
| 66 |
+
If you use our code or data in this repo or find our work helpful, please consider giving a citation:
|
| 67 |
+
|
| 68 |
+
```
|
| 69 |
+
@inproceedings{spec2024,
|
| 70 |
+
title={Synthesize Diagnose and Optimize: Towards Fine-Grained Vision-Language Understanding},
|
| 71 |
+
author={Peng, Wujian and Xie, Sicheng and You, Zuyao and Lan, Shiyi and Wu, Zuxuan},
|
| 72 |
+
booktitle={CVPR},
|
| 73 |
+
year={2024}
|
| 74 |
+
}
|
| 75 |
+
```
|