Detection and Tracking
a. Megvii has consecutively won the Object Detection task of COCO since 2017. The technology is applied to achieve high-speed and high-accuracy detection for all targets (face, pedestrian, vehicle, license plate) on all platforms（cloud, edge, device）
b. For full, high, and low frame rates, we provide a unique MOT(multiple object tracking) framework to self-adaptively balance the speed and the accuracy.
a. Megvii has consecutively won the Keypoints Detection task of COCO since 2017.
b. Supports 14, 17, 22 and 88 keypoints detection for 2D human body.
c. Supports 14 keypoints detection for 3D human body.
a.Megvii is the pioneer and leader of Re-Identification. We enabled the technology to be used in many industrial settings for the first time. We are at least one year ahead of our rivals.
b.Our Re-Identification is robust to any perspective, pose, illumination, and occlusion.
Attributes and Activity
a.Supports 100+ human attributes, including gender, age, hat, hair, clothes, bags, bicycles, umbrella, glasses, shoes, helmets, etc.
b.Supports human activities, such as running, falling, etc.
c.Supports adding a new attribute in 10 minutes by only a few samples.
License Plate Recognition
Supports the recognition of any license plate in any country.
Supports the recognition of 3000+ vehicle brands from any perspective.
Supports the 3D location of any vehicle and pedestrian from any monocular camera.