Facial Image Analysis

Facial recognition algorithms use deep neural network models helping smart devices to identify user, understand users’ characteristics, estimate users’ emotions, etc.

  • Face Recognition

    Support 1:1 face verification and 1:N face recognition, white/black list management

  • Face Detection and Tracking

    The system supports detecting and tracking human faces at real-time and is robust to rotation variations and occlusions.

  • Attributes Estimation

    Based on facial images, estimates attributes including age, gender, race. fast and accurate.

  • Liveness Detection

    The system supports both active and non-intrusive methods for liveness detection.

  • Facial Image Clustering

    Fast facial image clustering with locally computation. Suited for smartphone album management.

Human Sensing

Detects human from video streams, Understands human behavior and recognize human characteristics, Enable smart devices sensing the human presence and conducting tasks such as “human following”

  • Human Presence Detect

    Detect human presence in video stream with high accuracy

  • Human Detection

    detects and localize human with various poses in video stream

  • Human Tracking

    Tracks human movement. Can be resumed after discontinued by interrupts.

  • Human Movement Trajectories

    Detects human movement trajectories accurately in real-time. Supports both single and multiple person scenarios.

  • Hand Gesture Detection

    Detects and recognize hands. Enable smart devices understand static hand posture and dynamic gestures.

Environment Sensing

Detects obstacles using vision techniques, exploring an unknown environment and navigates. Enable smart devices navigates autonomously without users’ guidance

  • Vision Based Obstacle Avoidance

    Detects floor plane and obstacles, navigates autonomously. Help smart devices to sense obstacles more accurately.

  • Camera Positioning and Trajectory Tracking

    Vision-based SLAM, Detects and tracks features, Enable smart devices self-localization in unknown environments

  • Environment Understanding

    Combine 3D vision and deep learning, Estimate floor surface direction and object posture, Enable better understanding of the 3D space