The Datagen Humans in Context generator allows users to create datasets of people within different environments
Each sequence is a 10-second recording generated in 15-30 FPS, presenting a human interacting with a specific environment.
Acquiring high-quality face datasets for your computer vision system training is a huge headache. Gathering and cleaning this data is time-consuming, expensive, and riddled with privacy issues. For face datasets, the challenge is confounded by the need to account for limitless combinations of facial structures, poses, hairstyles, and expressions, not to mention a huge range of lighting conditions (including low light or nighttime environments), and camera angles.
The Datagen Platform provides high quality, perfectly annotated visual data in the form of video and images, with accurate representations of the in-cabin environment, advanced motion and human-object interactions for tasks related to driver action analysis, body keypoint estimation, gaze analysis, and hand pose analysis. Teams can use the platform to generate faces and full body simulated data to quickly iterate on their model and improve its performance.
Identity Scene – camera location, camera intrinsic matrix, lighting, and background
Detailed sequences created specifically for Driver Monitoring Systems, to explore computer vision tasks such as human detection, human pose estimation, attention analysis, and drowsiness detection.
Detailed sequences of conference rooms and home office environments, including multiple camera positions, tasks such as human detection: coming in/leaving a room and human pose estimation: standing up, sitting down, leaning forward, attention analysis and more.
Detailed sequences created specifically for human detection to explore computer vision tasks such as motion detection, pose estimation and object detection.
Detailed sequences of people performing various exercises in fitness environments including multiple camera positions, body keypoints estimation, pose analysis, posture analysis, repetition counting, object identification and more.
50K unique annotated identities, Granular control, on-demand generation.