johnolafenwa / DeepStack

The World's Leading Cross Platform AI Engine for Edge Devices
Apache License 2.0
678 stars 107 forks source link

Adds benchmark.py script #18

Closed robmarkcole closed 3 years ago

robmarkcole commented 4 years ago

Addresses https://github.com/johnolafenwa/DeepStack/issues/14 Placed in test directory as there are already images there. This shouldnt affect the test run as the filename doesn't begin with test.

robmarkcole commented 3 years ago

@johnolafenwa I want to improve this to also calculate an accuracy metric, something mAP@95. Any suggestions?

johnolafenwa commented 3 years ago

@robmarkcole , i will review the steps to add AP evaluating and revert back. It might be a little tricky but it is possible

johnolafenwa commented 3 years ago

@robmarkcole There are conflicts in merging this. Please review

robmarkcole commented 3 years ago

hmm followed the instructions but cannot see the conflict

robmarkcole commented 3 years ago

I actually now think this shouldn't be part of Dee-stack, but part of a separate utils library. Given this, and the conflict I cannot see, I will close this PR and we can have a discussion about where it should live

robmarkcole commented 3 years ago

Oh I see the merge conflict was because you already merged it, somehow

johnolafenwa commented 3 years ago

I see that, we can take out later and add it to the utils.