Drag100

The Drag100 dataset is introduced in the paper "GoodDrag: Towards Good Practices for Drag Editing with Diffusion Models"¹. This dataset is a new contribution to the benchmarking of drag editing¹.

The authors of the paper have developed this dataset to improve the evaluation and development of drag editing techniques¹. The Drag100 dataset, along with the proposed quality assessment metrics such as the Dragging Accuracy Index and Gemini Score, are utilized for extensive experiments¹. These experiments demonstrate that the proposed GoodDrag method compares favorably against the state-of-the-art approaches both qualitatively and quantitatively¹.

(1) GoodDrag: Towards Good Practices for Drag Editing with Diffusion Models. https://arxiv.org/abs/2404.07206v1. (2) Xiangyu Xu - GitHub Pages. https://xuxy09.github.io/. (3) GoodDrag: Towards Good Practices for Drag Editing with Diffusion Models. https://gooddrag.github.io/. (4) [2306.14435] DragDiffusion: Harnessing Diffusion Models for Interactive .... https://arxiv.org/abs/2306.14435. (5) undefined. https://doi.org/10.48550/arXiv.2404.07206.