Open jjuod opened 2 years ago
Also, additional/better diagnostics could be produced: some summary statistics about the temporal detection precision (average distance between detected and marked endpoints etc.), or data for classification performance vs. SNR (already have functions to calculate SNR without resorting to manual noisiness labelling).
Filter testing can take a long time and currently blocks the GUI thread, i.e. appears unresponsive. It should use a separate thread. Could use the same threading mechanics as batch processing mode - testing uses the same
batchProcess
worker already, so should be straightforward.Note: don't be tempted to do the same for the CLI mode, as it should not use any Q* things including QThreads. No reason to multithread CLI anyway.