search

dbolkensteyn / BeatTheBotChallenge

Come challenge the bot playing a moto racing phone game, see how it is made, and help to improve it!
3 stars 2 forks source link

Static detector: Use a bigger image database #18

Closed dbolkensteyn closed 9 years ago

dbolkensteyn commented 9 years ago

Fixed in https://github.com/dbolkensteyn/BeatTheBotChallenge/commit/fab1c283e9c67b58f22ec7bb15531484d7d6844b

dbolkensteyn commented 9 years ago

As a consequence of this bigger training database, we are able now to detect more motos (~50% in the StaticDetectorTest) - but we also now start to detect a lot of "garbage", and are at 40% of mismatching rate:

Matches: 38, Mismatches: 29 (average distance: 48.0261), Total: 75,
Matching rate: 50.6667%, Mismatching rate: 38.6667%

The idea is that, with more training images, more "noise" keypoints are also added, which will be successfully matched against scene images - yielding garbage results. Issue #19 has been created to try to detect and prune these weak keypoints.