I would like to thank the orgnizers, sponsors and driven data staff for this amazing competition.
And congratulation to all winners.
Just a quick update:
My last submission which is Eligible and meet the rules scored 0.4887 on public and 0.4954 on private which I guess will keep me on place.
The main problem was that I transferred the literal top into top problem only and the models which trained for literal top were trained to maximize the top1 accuracy. So on the scenarios where the query images are top and the database are literal I was using my top models to find top2 from other queries and add the similarities which was not elligable (but I didn’t know that :() so I just removed that on my last submission (good for me that I just noticed before sending the last submission).
Thanks it was a great honor to be a part of this ))
Congratulations, well done
1 Like
Oh thanks and happy birthday )). You did an amazing job on this competition as well congrats. I don’t know how qwerty64 just jumped like that on 3 submissions. I was like oh I will get a heart attack when he passed me on the public LB 