Learning video retrieval models with relevance-aware online mining

TL;DR

This paper introduces Relevance-Aware Negatives and Positives mining (RANP), a novel method for improving cross-modal video retrieval by better selecting training samples based on semantic relevance, leading to state-of-the-art results.

Contribution

The paper proposes RANP, a new technique for selecting negatives and positives in training video-text retrieval models, addressing the issue of wrongly penalizing valid positives.

Findings

Achieves +5.3% nDCG on EPIC-Kitchens-100

Achieves +3.0% mAP on EPIC-Kitchens-100

Improves retrieval performance by better sample mining

Abstract

Due to the amount of videos and related captions uploaded every hour, deep learning-based solutions for cross-modal video retrieval are attracting more and more attention. A typical approach consists in learning a joint text-video embedding space, where the similarity of a video and its associated caption is maximized, whereas a lower similarity is enforced with all the other captions, called negatives. This approach assumes that only the video and caption pairs in the dataset are valid, but different captions - positives - may also describe its visual contents, hence some of them may be wrongly penalized. To address this shortcoming, we propose the Relevance-Aware Negatives and Positives mining (RANP) which, based on the semantics of the negatives, improves their selection while also increasing the similarity of other valid positives. We explore the influence of these techniques on two video-text datasets: EPIC-Kitchens-100 and MSR-VTT. By using the proposed techniques, we achieve considerable improvements in terms of nDCG and mAP, leading to state-of-the-art results, e.g. +5.3% nDCG and +3.0% mAP on EPIC-Kitchens-100. We share code and pretrained models at \url{https://github.com/aranciokov/ranp}.