trackpy.predict.instrumented

trackpy.predict.instrumented(limit=None)

Decorate a predictor class and allow it to record inputs and outputs.

Use when diagnosing prediction.

limit : maximum number of recent frames to retain. If None, keep all.

Examples

>>> pred = instrumented()(ChannelPredict)(50, flow_axis='y')
>>> pred.link_df_iter(...)
>>> diagnostics = pred.dump()