Datasets:

Modalities:
Image
Size:
< 1K
Libraries:
Datasets
License:
File size: 917 Bytes
7382154
 
cd117af
 
280469e
db27d59
cd117af
 
7382154
cd117af
 
 
7d4cc3a
682669f
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
---
license: apache-2.0
task_categories:
- image-to-image
- text-to-image
pretty_name: TEdBench++
size_categories:
- n<1K
---

# TEdBench++

This dataset contains the TEdBench++ an image-to-image benchmark for text-based generative models. It contains original images (originals) and edited images (LEdits++) for benchmarking. ``tedbench++.csv`` contains the text-based edit instructions for the respective original image and parameters to reproduce the edited images with LEdits++.

consider citing our work

```bibtex
@inproceedings{brack2024ledits,
  year = { 2024 },
  booktitle = { Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) },
  author = { Manuel Brack and Felix Friedrich and Katharina Kornmeier and Linoy Tsaban and Patrick Schramowski and Kristian Kersting and Apolinaros Passos },
  title = { LEDITS++: Limitless Image Editing using Text-to-Image Models }
}
```