Open abuzarmahmood opened 8 months ago
Need code to generate plots after inference. Starting point would be to generate plots of the following kinds as well as:
Plot showing timeseries of blocks of mouth movements colored by what movement they were.
First draft of plots
Currently too many false positives because classifier doesn't have a negative category. Can use MAD threshold on baseline to mark negative events. Likely will be better to use top-hat filtered data to avoid long deflections as seen in trials #7 and 8.
We should also probably train classifier on top-hat filtered data because the first deflection in trial #16 above is being registered as a gape (see overlay lineplot in comment further above).