In the field of machine learning and deep learning, datasets plays an irreplaceable role. No matter it is image data for convolutional neural networks or massive text data for natural language processing, the integrity and diversity of data directly determine the learning results of a model. With the advancement of technology, datasets that collected from specific scenarios have becomes the core strategy for improving model performance.

Gesture recognition in smart home scenario can realize the remote control of smart TVs, smart air conditioners, and smart speakers, according to the coordinate information of the hand or body.

In the field of intelligent driving, gesture recognition is also a hot human-computer interaction method in recent years. Automobile manufacturers increasingly hope to reduce distractions of the driver by implementing functional safety technologies in their cars, so that drivers can drive safely.

Gesture recognition technology has undergone continuous improvement and innovation before it can be applied in multiple scenarios. In general, the development of gesture recognition technology is a process from static to dynamic, from two-dimension to three-dimension.

According to the gesture state, gesture recognition can be divided into static gesture recognition and dynamic gesture recognition. Static gesture recognition is for a single hand shape, and mainly includes two parts: gesture segmentation and gesture recognition, the former is the basis of the latter. Static gesture recognition is currently very mature technology, but dynamic gesture recognition can recognize gestures with less discrimination and is more fault-tolerant. Therefore, the transition from static gesture recognition to dynamic gesture recognition has become a trend.

As an AI data service provider that is deeply involved in AI data field for ten years, Nexdata is committed to providing professional data services to global artificial intelligence companies. We have launched high-quality static and dynamic gesture recognition training data to support the innovation in gesture recognition technology.

314,178 Images 18 Gestures Recognition Data

Collection environment : indoor scenes and outdoor scenes (natural scenery, roadside street view, square, etc.)

Data diversity : multiple scenes, 18 gestures, 5 shooting angels, multiple ages, multiple light conditions

Accuracy : the accuracy of gesture type and gesture attributes are not less than 95%

180,718 Images — Sign Language Gestures Recognition Data

Data size : 180,718 images, including 83,013 images of static gestures, 97,705 images of dynamic gestures

Collection environment : including indoor scenes and outdoor scenes

Accuracy : accuracy requirement: the point location errors in x and y directions are less than 3 pixels, which is considered as a qualified annotation; accuracy of landmark annotation: the annotation part (each landmark) is regarded as the unit, the accuracy rate shall be more than 95%.

559,460 Videos — 50 Types of Dynamic Gesture Recognition Data

Data size : 559,460 videos, 220,030 videos were collected by laptop, 339,430 videos were collected by cellphone or iPad

Collection environment : including indoor scenes and outdoor scenes (natural scenery, street view, square, etc.)

Accuracy : based on the accuracy of the gesture actions, the accuracy exceeds 97%; the accuracy of the video naming exceeds 97%; the accuracy of metadata file annotation exceeds 97%

End

If you need data services, please feel free to contact us：[email protected]

With the advancement of data technology, we are heading towards a more intelligent world. The diversity and high-quality annotation of datasets will continue to promote the development of AI system, create greater society benefits in the fields like healthcare, intelligent city, education, etc, and realize the in-depth integration of technology and human well-being.