=============

This code was using for increasingly popular gesture recognition trained on CSI（Channel State Information,信道状态信息） data.

Our method is based on Deep Metric Learning and choosing CNN/ ResNet/ VGG as backbone network. We conduct and evaluate this three types of network with different structure on SignFi Dataset. However, for better results, we make some augmentation on the SignFi Dataset and shared is here, the password is "5xe8".

We call the result Network structure CSI-Gesture-Recognition-Network.

2020/6/27

This work is done for IoT lesson on 2020 spring term, Sun Yat-sen University. For further improvement, we can implement SVMs or Softmax Function as the state-of-the-art classifier to tackle the Feature Map output by our proposed CSI-Gesture-Recognition-Network.