Topographic Design in Wearable MXene Sensors with In-Sensor Machine Learning for Full-Body Avatar Reconstruction
This is the repo for the project Topographical Design in Wireless Strain Sensor for Deep Learning-Enabled Whole-Body Motion Tracking
This repo contains the code for analysing the Gn sensor data during various human motions, including the code for ANN model training to realize the motion classification, the code for t-distributed Stochastic Neighbor Embedding (t-SNE) to visulize the Gn sensor data, as well as the code for CNN model training for full-body avatar reconstruction.
After the images are predicted and drawed by our codes, we used https://imgflip.com/gif-maker to create the gif file
The motion classification process for the ANN model is done by the ANN_sensor_classification.py and we also have the google colab link as in https://colab.research.google.com/drive/1YY4BdhhYE6sSNL_A3zg2GegObW35JlQ6.
The major model training process for the CNN model is done by the CNN_model_trainging.py and we also have the google colab link as in https://colab.research.google.com/drive/1Su69f_3Y_L7QD7jItklj3BN8R35NyTr3?usp=sharing
Supporting files of Tables S2–S6 in the manuscript are in Supporting-Tables folder