19 datasets found

UniEdit: A Unified Tuning-Free Framework for Video Motion and Appearance Editing

The dataset used for speech-to-speech translation, lip-syncing, and talking face generation.

VideoMap is a proof-of-concept video editing interface that operates on video frames projected onto a latent space, enabling users to visually uncover patterns and relationships.

The dataset used for training the InstructVid2Vid model, which consists of video-instruction-edited video triplets.

The VASE dataset is used for video object-centric appearance and shape edits. It consists of 30 videos from the DAVIS dataset, with 3 driver images to edit each video.

The dataset used in the paper is not explicitly described, but it is mentioned that the authors collect 3K videos (60-90 seconds long) from the internet to train their model.

Text-based video editing using MaskINT, a two-stage pipeline involving keyframe joint editing and structure-aware frame interpolation.

The Emu Video Edit model's training dataset, containing 1600 videos with 7 editing instructions each.

The dataset used for training the Emu Video Edit model, containing 1600 videos.

Zero-shot text-based video editing with depth map guidance and temporal consistency constraints

Make-a-protagonist: Generic video editing with an ensemble of experts.

Zero-shot video editing using off-the-shelf image diffusion models.

ControlVideo is a general framework to utilize T2I diffusion models for one-shot video editing, which incorporates additional conditions such as edge maps, the key frame and...

The dataset used in the paper for video editing tasks

The DAVIS dataset is a widely used dataset for video-related tasks, consisting of approximately 2000 frames from 26 human-centric scenarios.

The dataset used in the paper is not explicitly described, but it is mentioned that the authors used the DAVIS dataset and the Pexels website to gather face videos.

The Show-1 dataset is a text-to-video diffusion model.

The dataset used in this paper for text-to-video generation, consisting of short video clips.

The dataset used in the paper is not explicitly described, but it is mentioned that the authors used 26 text-video pairs from the public DAVIS and WebVid datasets.