HaGRID (HAnd Gesture Recognition Image Dataset)
收藏资源简介:
HaGRID是由俄罗斯的SaluteDevices创建的一个大规模手势识别图像数据集,包含554,800张图像,用于构建手势识别系统,特别关注与设备交互以管理它们。数据集中的18种选定手势具有符号功能,可解释为特定动作。尽管手势是静态的,但它们被特别挑选出来,以便能够设计几种动态手势。该数据集通过利用众包平台收集,涵盖了从0.5到4米不等的相机到主体距离,以及各种自然光照条件。HaGRID不仅适用于静态手势识别,如“喜欢”和“停止”,还适用于“滑动”和“拖放”等动态手势。数据集的应用领域包括家庭自动化设备和视频会议服务,旨在通过直观和简单的手势提高用户体验和加速过程。
HaGRID is a large-scale gesture recognition image dataset developed by SaluteDevices, a Russian company, containing 554,800 images. It is designed for building gesture recognition systems, with a special focus on device interaction for device management. The 18 selected gestures in the dataset have symbolic functions that can be interpreted as specific actions. Although these gestures are static, they are specifically chosen to enable the development of several dynamic gestures. This dataset was collected through crowdsourcing platforms, covering camera-to-subject distances ranging from 0.5 to 4 meters and various natural lighting conditions. HaGRID supports not only static gesture recognition tasks such as "like" and "stop", but also dynamic gesture recognition scenarios like "swipe" and "drag-and-drop". Its application fields include home automation devices and video conferencing services, aiming to improve user experience and accelerate workflows via intuitive and straightforward gestures.




