UniUD-FBK-UB-UniBZ Submission to the EPIC-Kitchens-100 Multi-Instance Retrieval Challenge 2022
This is an incremental improvement for video retrieval in kitchen environments, addressing a specific challenge benchmark.
The authors tackled the EPIC-Kitchens-100 Multi-Instance Retrieval Challenge by designing an ensemble of models trained with relevance-augmented triplet loss variants, achieving an average score of 61.02% nDCG and 49.77% mAP.
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.