< DEMO > My work and DEMO

Blinking DEMO

The blue digits on the right top of the video is the time of blinks since detection started.

For more details, you can check this page.

FaceAntispoofing, RNN Smooth DEMO

The number behind the string “CNN” is the output of CNN model which is used to predict a face whether is fake or alive.

The number behind the string “RNN” is the output of RNN model which aims to smoth the final result fed by CNN model based on sequence.

The number behind the string “SE_NUM” is the number of sequence which is used by RNN model to smooth the final result.

For details of RNN, you can check this page .

Face Antispoofing, RNN online Training DEMO

When white string “DO” is displayed on the top of the video, it meams that the RNN model is inferencing without any training and the number behind the string “RNN” is the confidence output of RNN model which is used to predict a face is whether fake or alive.

When white string “TRAIN_LIVE” is displayed on the top of the video, it meams that RNN model are trained online only based on liveness samples captured by camera.

When white string “TRAIN_ATTACK” is displayed on the top of the video, it meams that RNN model are trained online only based on spoofing samples captrued by camera.

The number behind the string “LOSS” is the current loss while training model online.

For details of RNN online training, you can check this page .

RNN based lane line offset detection DEMO

The number on the bottom left of the video indicates the probability whether a car is about to changing a lane or not.

Face Antispoofing Attacking DEMO

For more details, please check this page