Question

I’m using python to prototype the algorithms of a computer vision system I’m creating. I would like to be able to easily log heterogeneous data, for example: images, numpy arrays, matplotlib plots, etc, from within the algorithms, and do that using two keys, one for the current frame number and another to describe the logged object. Then I would like to be able to browse all the data from a web browser. Finally, I would like to be able to easily process the logs to generate summaries, for example retrieve the key “points” for all the frame numbers and calculate some statistics on them. My intention is to use this logging subsystem to facilitate debugging the behaviour of the algorithms and produce summaries for benchmarking.

I’m set to create this subsystem myself but I thought to ask first if someone has already done something similar. Does anybody know of any python package that I can use to do what I ask?
otherwise, does anybody have any advice on which tools to use to create this myself?

Answer 1

Another option for storage could be using hdf5 or pytables. Depending on how you structure the data, with pytables you can query the data at key “points”. As noted in comments, I dont think an off the shelf solution exists.