80 datasets found

A dataset consisting of 11 categories of action videos, which are recorded under dark environments.

NTU-RGBD 120 is an extension of NTU-60 with additional 57,367 skeleton sequences, which totally contains 114,480 samples.

NTU-RGBD 60 is one of the widely used datasets for skeleton-based action recognition tasks, which contains 56,880 samples of 60 different classes.

Skeleton-based action recognition task has received significant attention due to its computation efficiency and robustness against viewpoints or appearance.

The dataset used in the paper is Something-Something V2 for action recognition.

The dataset used in the paper is Kinetics400 for action recognition.

The UCF101-24 dataset is a subset of the UCF101 dataset, containing 3207 videos with spatio-temporal annotations on 24 action categories.

The HMDB dataset consists of 6766 videos from 51 different action categories. The videos are generally of low quality, with strong camera motion, and non-centered people.

The AVA v2.2 dataset for spatiotemporal action localization contains the bounding box annotations and the corresponding action labels on keyframes.

A large-scale human motion capture dataset with unique action categories.

Generating human action proposals in untrimmed videos is an important yet challenging task with wide applications. Current methods often suffer from the noisy boundary locations...

The High-Quality Fall Simulation Dataset (HQFSD) is a challenging dataset for human fall detection, including multi-person scenarios, changing lighting, occlusion, and...

UCF-24 and JHMDB-21 are two public action datasets used for evaluation of action detection algorithms.

The Mini-Kinetics dataset is a mini version of the Kinetics-400 dataset, containing 240k training samples and 20k validation samples in 400 human action classes.

NW-UCLA is a small-scale dataset commonly used in this field to test the robustness of methods, including 1494 video clips with 10 action categories performed by 10 different...

Skeleton-based action recognition has attracted considerable attention due to its compact representation of the human body’s skeletal structure. Many recent methods have...

A sports video highlight summarization framework using YOLO v3 and OpenPose.

The HMDB51 dataset is a video dataset for human action recognition. It contains 6,767 videos annotated with 51 categories of human actions.

Ego4d is a large-scale egocentric video dataset that contains 3000 hours of video.

Skeleton-based action recognition has attracted considerable attention due to its compact representation of the human body’s skeletal structure. Many recent methods have...