aosokin / os2d

OS2D: One-Stage One-Shot Object Detection by Matching Anchor Features

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

New form of evaluation

saswat0 opened this issue · comments

Reusing this codebase, can we get the prediction and ground truths in two text files in the following format?
<class_name> <left> <top> <right> <bottom> [<difficult>]

I think the input file (e.g., classes/grozi.csv from here) is in a format extremely similar to this one. I don't think we ever implemented export in this format but that should be easy to do by saving a file at the end of the evaluation function.

Thanks. That's helpful.
Any idea on how to do the same with the baselines as well?

Sorry, I totally do not remember the details about the baselines anymore :-(