Computer Vision Datasets

This dataset includes 2,255 images of 451 people across multiple races. The semantic segmentation area includes headphones, glasses, body and background.This dataset can be used for training AI models in human body segmentation, video conference behavior detection, and smart vision applications.

This 3D high-fidelity synthetic dataset simulates real-world Driver Monitoring System (DMS) environments using photorealistic 3D scene modeling. It includes multi-modal sensor outputs such as camera images, videos, and point clouds, all generated through simulation. The dataset is richly annotated with object classification, detection, and segmentation labels, as well as human pose data (head, eye, arm, and leg position/orientation), camera parameters, and temporal metadata such as illumination and weather conditions. Ideal for training and evaluating models in autonomous driving, robotics, driver monitoring, computer vision, and synthetic perception tasks.

The Driver Gesture Recognition Dataset contains recordings from 1,334 participants across diverse age groups, time periods and multiple gestures. Each person performed 18 static gestures(such as fist-clenching gestures and heart-to-heart gestures) and 23 dynamic gestures(including index finger clicks and two-finger clicks). In terms of acquisition equipment, visible light and infrared binocular cameras are used. This dataset is ideal for tasks such as driver monitoring, gesture recognition in vehicles, human-computer interaction, and driver assistance systems.

415 People Living Face & Anti-Spoofing Dataset – Multi-Scene, Multi-Expression. The collection scenes include indoor and outdoor scenes. The data includes male and female. The age distribution ranges from juvenile to the elderly, the young people and the middle aged are the majorities. The data includes multiple postures, multiple expressions, and multiple anti-spoofing samples. The data can be used for tasks such as face payment, remote ID authentication, and face unlocking of mobile phone.

This dataset contains facial images from 6,502 individuals, with 7 images collected for each person under diverse real-world conditions. The dataset covers various demographic characteristics, including different age groups and ethnic backgrounds,covering Asian, Black, Caucasian, and other populations, as well as multiple environments such as indoor and outdoor scenes. It includes different mask types, facial occlusion conditions, and lighting variations. The dataset is suitable for masked face recognition, face verification and face identification.

This dataset contains face images from 21,125 individuals with diverse demographic backgrounds(Southeast Asian, Caucasian, Black, Brown and Indian races.). Each person was captured with 29 images under different poses, lighting conditions, including 28 variation images and one identity reference image. The dataset is suitable for face recognition, face verification, identity matching, face detection, and other computer vision applications.

This dataset collected from 2,241 people in online conference scenarios, Participants include Southeast Asians, Caucasian, Black, Brown individuals, mainly young and middle-aged people. Data was collected from a variety of indoor office scenes, covering meeting rooms, coffee shops, library, bedroom, etc. Each participant contributed 23 videos covering actions like opening the mouth, turning the head, closing the eyes, and touching the ears, with 4 images featuring variations such as wearing a mask and sunglasses.This dataset is suitable for tasks such as human activity recognition, human action recognition, human pose estimation, gesture recognition, mask detection, and AI applications in video conferencing and virtual meetings.

This dataset collected from 2,241 people in online conference scenarios. Participants includes Asians, Caucasians, blacks, and browns. The age is mainly young and middle-aged. It collects a variety of indoor office scenes, covering meeting rooms, coffee shops, libraries, bedrooms, etc. Each person collected 11 videos, including human body behaviors such as shaking the body from side to side, eating, and stretching. This dataset can be applied to tasks including human activity recognition, human action recognition, human behavior analysis, human pose estimation, and gesture recognition.

This dataset contains facial expression recognition data from 1,130 people in online conference scenes. Participants include Asian, Caucasian, Black, and Brown individuals, mainly young and middle-aged adults. Data was collected across a variety of indoor office scenes, covering meeting rooms, coffee shops, libraries , bedroom, etc., Each participant performed seven key expressions: normal, happy, surprised, sad, angry, disgusted, and fearful. The dataset is suitable for tasks such as facial expression recognition, emotion recognition, human-computer interaction, and video conferencing AI applications.