53 datasets found

YouCook2 consists of recipes containing labels that separate the long horizon trajectories of demonstrations into events - with explicit time stamps for the beginning and end of...

TimeIT is a video-centric instruction-tuning dataset designed for instruction tuning. It is composed of 6 diverse tasks, 12 widely-used academic benchmarks, and a total of 125K...

TimeChat is a time-sensitive multimodal large language model specifically designed for long video understanding. It incorporates two key architectural contributions: a...

Face aging techniques have used generative adversarial networks (GANs) and style transfer learning to transform one’s appearance to look younger/older. Identity is maintained by...

The dataset used in the paper for the SemEval-2021 task 6: Detection of persuasion techniques in texts and images using CLIP features.

The Reuters Video-Language News Dataset (ReutersViLNews) is a large-scale video-language understanding dataset containing 1,974 long-form news videos with an average video...

The Hateful Memes Dataset consists of a training set of 8500 images, a dev set of 500 images & a test set of 1000 images. The meme text is present on the images, but also...

The OK-VQA dataset is a visual question answering benchmark requiring external knowledge.

Unpaired vision-language pre-training via cross-modal CutMix.

Multimodal sentiment analysis (MSA) leverages heterogeneous data sources to interpret the complex nature of human sentiments.

InternLM2 is a vision-language large model that supports images with any aspect ratio from 336 pixels up to 4K HD, facilitating its deployment in real-world contexts.

FocusCLIP: Multimodal Subject-Level Guidance for Zero-Shot Transfer in Human-Centric Tasks. This paper introduces FocusCLIP, an enhancement for CLIP pretraining using a new ROI...

QVHighlights is a dataset for video highlight detection, which consists of over 10,000 videos annotated with human-written text queries.

The WavCaps dataset contains chatGPT-assisted weakly-labeled audio captioning data.

The Multimodal Visual Patterns (MMVP) benchmark is a dataset used to evaluate the visual question answering capabilities of multimodal large language models (MLLMs).

Multimodal C4 (mmc4) is a public, billion-scale corpus of images and text, constructed from public webpages contained in the cleaned English c4 corpus.

Video question answering (VideoQA) requires systems to understand the visual information and infer an answer for a natural language question from it.

The MLM dataset is a collection of images and captions that represent different cultures from around the world.

Stanford Large Movie, Games and Datasets Archive (SMLMDA) dataset is used for training and evaluation.

The dataset used in the paper is a collection of pairs of observations (xi, ˜xi) from two modalities, where xi ∈ Rd1 and ˜xi ∈ Rd2. The dataset is used to evaluate the...