UniUD-FBK-UB-UniBZ Submission to the EPIC-Kitchens-100 Multi-Instance Retrieval Challenge 2022

22 Jun 2022  ·  Alex Falcon, Giuseppe Serra, Sergio Escalera, Oswald Lanz ·

This report presents the technical details of our submission to the EPIC-Kitchens-100 Multi-Instance Retrieval Challenge 2022. To participate in the challenge, we designed an ensemble consisting of different models trained with two recently developed relevance-augmented versions of the widely used triplet loss. Our submission, visible on the public leaderboard, obtains an average score of 61.02% nDCG and 49.77% mAP.

PDF Abstract

Datasets


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Multi-Instance Retrieval EPIC-KITCHENS-100 UniUD-FBK-UB-UniBZ mAP(V2T) 55.15 # 1
mAP(T2V) 44.39 # 1
mAP (Avg) 49.77 # 1
nDCG (V2T) 63.16 # 1
nDCG (T2V) 58.88 # 1
nDCG (Avg) 61.02 # 1

Methods


No methods listed for this paper. Add relevant methods here