About Simple Eye

Simple Eye is an innovative laboratory based in Nanjing, Beijing, Shanghai and Silicon Valley, which specializes and focuses on computer vision and machine learning technologies. We develop advanced algorithms of Face Alignment, Face MakeUp, Face Swap, etc, and implement highly optimized codes for embedded platforms.


Face Alignment

Face Alignment technology can achieve accurate detection and tracking up to 66 facial landmarks under complex and unconstrained environment. Facial landmarks represent the key positions on face, such as eye center, nose tip, mouth corner, etc. Our machine learning and big data based face alignment algorithm is capable of accurately detecting facial key points under various expressions and poses of face.


Creating real-time face makeup effect on the video with the prototype of another style example image.

Our technology is analogous to the physical makeup process, and it can effectively transfer the special cosmetics appearance, such as foundation, powder, cream, rouge, eye liner and shadow to the target user face under any pose and expression, while preserving the original face structure.

Face Swap

Facial replacement plays a crucial role in human interaction. Our face swap technology for mapping and manipulating facial expression in real-time across images and video streams, facial expression migration and image fusion technology based on facial feature point localization .Our system is fully automatic, and takes advantage of several computationally inexpensive methods to work in real-time, on generic mobile phones.

Face Video Beautification

Automatic face photo/video retouching technology aiming to enhance the attractiveness of the user.

Our algorithm can not only detect and remove flaws, moles, and acne from faces through region-aware manipulations, but also improve facial lighting and adjust the face structure to enhance the attractiveness. The performance of our algorithm is stable under unexpected situations and the computational requirement is low.

Facial Animation

Real-time avatar facial expression animation driven by the user performance on mobile devices.

Our single-camera based facial animation algorithm does not rely on any special equipment (e.g., facial markers and structure light projectors). User facial parameters, including head poses and expression coefficients, are estimated in real time from 2D user video frames. These facial parameters are then transferred onto the target still face image for the effect of animation. The target image can be a face photo as well as a face cartoon image.

Cartoon Face Generation

Real-time automatic cartoon face generation with various cartoon styles, such as Gray-Scale cartoon, Ink-Drawing style etc.

Based on several standard image templates, our face/off algorithm is able to implement the real-time automatic facial image fusion. Our algorithm is available either with a single picture or with taking picture in real time.

Visual Object Tracking

Visual object tracking is a technology to recover the trajectories of respective objects in a given video. It plays a key role in many intelligent visual systems, such as video surveillance, identity recognition, and behavior recognition.

Face Parsing

Real-time face labeling under complex face poses and visual patterns. Face labeling is to parse an input face into facial parts, eg. eyes and hair. Our machine learning based algorithm is capable of accurately labeling facial parts under various face pose and complex environment.