自動車の外観で車種を識別することができるのか？ - VW Tracking Contest 2013 - Szenario 2
自動車の外観で車種を識別することができるのだろうか？従来のAR認識技術では、不可能かまたは完全とは言えないものが多かった。このビデオでは、ARの認識をフォルムと特徴点の２つの方法を併せて使用しているように見える。この方法に更に対してさらに色データを検出対象にできたらより精度が高まるだろう。PCクラスのグラフィックチップと高性能省電力CPUがスマートフォンに組み込めるようになったら実用に耐えうるだろう。
Scenario two is about the transfer of tracking data used during the (offline) learning phase and (online) tracking. The contest goal is to precisely locate and track points on the exterior/surface of a car. This is tricky, because the car from the preparation and the run were of the same model, but the latter had a different color and interior components (i.e. the cars were similar but not exactly the same).
http://tracking-challenge.de/content/...
(via https://www.youtube.com/channel/UC6lUnpTfqiIQqLsy31RqfCg)
Scenario two is about the transfer of tracking data used during the (offline) learning phase and (online) tracking. The contest goal is to precisely locate and track points on the exterior/surface of a car. This is tricky, because the car from the preparation and the run were of the same model, but the latter had a different color and interior components (i.e. the cars were similar but not exactly the same).
http://tracking-challenge.de/content/...
(via https://www.youtube.com/channel/UC6lUnpTfqiIQqLsy31RqfCg)
