74 datasets found

The AVA database contains over 250,000 images with multiple aesthetic scores from reviewers.

YouTube-8M is a large-scale video classification benchmark.

Oxford RobotCar dataset contains a large amount of data collected from one route through central Oxford, and covers various weather and traffic conditions.

The dataset used in this paper is a large-scale dataset for automatic detection of cash-out fraud, with more than 100 million training samples.

The Open Image Dataset (OID) is a large-scale image dataset that contains a diverse set of images.

The dataset used in the paper is not explicitly described, but it is mentioned that the authors used a few-shot relation extraction task (TACRED) and a few-shot variant of TACRED.

A novel large-scale dataset for face forgery detection in multi-person scenarios, comprising 10,000 high-quality forgery videos with an average of three human faces in each frame.

N-UCLA dataset is a widely used skeleton-based action recognition dataset, containing 1494 video clips featuring 10 volunteers.

A large-scale dataset of human poses and their text descriptions extracted from large-scale image datasets.

The dataset is used for learning general and effective models for both audio and video analysis from self-supervised temporal synchronization.

LRS3-TED: a large-scale dataset for visual speech recognition.

A large-scale multimodal machine translation dataset named How2, which has 1.57 times longer mean sentence length than Multi30k and no repetition.

The YFCC100M dataset.

The ImageNet-2012 training set of 1.2 million images labelled into 1,000 object categories.

The LAION-Aesthetics dataset is a large-scale dataset of images used for training and evaluating computer vision models.

The MIT-Adobe FiveK dataset, a large-scale dataset for image segmentation and object detection.

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

A large-scale multi-modal dialogue dataset created by leveraging the automatic pipeline with filtering using CLIP similarity.

The dataset used in the experiment consists of 50 images equally distributed between five classes: aircraft, bird, bicycle, boat, and dog. Each class has 5 true positive images...

The LAION-400M dataset is a large collection of images and captions that represent different cultures from around the world.