ANT Group’s AI technology specializes in image recognition and image processing products. The face recognition algorithm identifies facial features by extracting boundaries, or traits, from a face image of an object. The algorithm then extracts the information, and these features are then used to search for other images with matching features. The algorithm simplifies a set of face images. and then compress face data, only saving image data which is useful for face recognition.

ANT Group uses biometric face recognition technology with many advantages such as being able to operate in real time, using more than 80 different properties of faces, high accuracy through mass comparision and processing. With distributed pre-processing capability it helps to reduce the load on the central server.

The system uses face recognition technology to automatically check attendance and track the attendance time of each student.

Main applications:

– Attendance at the beginning of class
– Keep track of students’ time of attendance

Working mechanism:

Monitor the environment:
– Applying fast processing algorithms such as Back ground Subtraction to rough assess the fluctuation of environment in the class. Advantages of fast processing speed, little calculation, help reduce the load on peripheral computers

Detect people:
– Use high-accuracy human detection models like Faster R-CNN to detect people moving in class. Model needs large computational performance on the server

Tracking object:
– Based on the location of detected objects, object tracking models based on image processing or Deep Learning-based models (such as dlib) are used to track the object’s movement.

Face Registration:
– Use the tool (web) to take photos of each student’s face registration before using the system

Building an Automatic Sorting System, used to classify products with high accuracy to reduce time and cost

Based on the product sample data set (standard and error), the model learns from the data and then automatically identifies the features and automatically finds a set of parameters that help the system to classify product quality.

Compared to systems using pure image processing technology, the complexity of the system using AI is much higher. It is possible to classify complex cases, which require a lot of conditions to classify. ) Works well on a variety of environmental conditions (e.g. brightness varies)