I have got multi-class object detector. One model accuracy evaluation of detection consists of: mAP, FP, FN, TP for each class divided to two graphs and looks like this (I've used this repo for evaluation).
Now, I've got many of these evaluations (multiple times these two graphs for different models) and I would like to easily compare all these trained models (results) and put them to one graph.
I've searched through the whole Internet, but wasn't able to find suitable method of placing all the values to one graph. Also, the values of these three classes can be put together (eg. result mAP for this evaluation would be (75+ 68+ 66) / 3 = ~70%), so I would have just single value of each mAP, FN, FP, TP for one whole model evaluation.
What comes to my mind is the following graph (or maybe some kind of plot):
Note: It may not make sense to place mAP together with TP, etc. into one graph, but I would like to have all these values together to easily compare all the model evaluations. Also I am not really looking for a script, I can do the graph manually from values, but script would be more helpful. What really matters is, how to create meaningful graph with all the data :). If the post is more suitable for different kind of site, please, let me know.