Prediction visualisation can be found in [reports\report.md](reports\report.md) file

Good result for 30 gausses is on the 46c92265c2eb15eaa74491cf3580919414ff53cf commit.

Table for comparing results and quality of prediction is [here](https://studio.iterative.ai/user/korbash1/views/RQC-Robotics-tactile_sensor-vvk6t1pklh?mode=full)

install

The easiest way to run our code is using Google Colab. You need to load the data folder to your Google Drive.
You can download data here:  https://drive.google.com/drive/folders/1qfujkRPA81V8XF0RwKlAvepwT75hBL8Y?usp=sharing
Then you should open examples in Colab. 
main.ipynb is example in which:
1) generated pressure profiles (you can skip this step because it take too much time)
2) simulated output of our sensor for each input profile.
3) training NN for decoding input profile by the simulated output. (for our configuration you need to have colab pro)
4) saving model prediction 

See_results.ipynb is an example with a some result analysis. 
You can check predictions of our net for different sensor structure. 
And look to the worst, the best and medium predictions.

In both notebooks you need to put in variable dir_of_data the part to folder with data on your Google Drive.