Exploring Phrase Grounding without Training: Contextualisation and Extension to Text-Based Image Retrieval

Pârcălăbescu, Letiția; Frank, Anette

doi:10.1109/cvprw50498.2020.00489

Cited by 5 publications

(1 citation statement)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since plural phrases are much less frequent than singular phrases, this distributional bias may lead to poorer predictions for plural phrases. Recently, unsupervised PG systems have been proposed (Wang and Specia, 2019;Parcalabescu and Frank, 2020) that achieve competitive performance, but are not subject to such frequency biases. We thus perform our experiments with a system that replicates Wang and Specia (2019)'s approach.…”

Section: Bias In Context Of System Performancementioning

confidence: 99%

Grounding Plural Phrases: Countering Evaluation Biases by Individuation

Suter¹,

Pârcălăbescu²,

Frank³

2021

Proceedings of the Second Workshop on Advances in Language and Vision Research

Self Cite

View full text Add to dashboard Cite

Phrase grounding (PG) is a multimodal task that grounds language in images. PG systems are evaluated on well-known benchmarks, using Intersection over Union (IoU) as evaluation metric. This work highlights a disconcerting bias in the evaluation of grounded plural phrases, which arises from representing sets of objects as a union box covering all component bounding boxes, in conjunction with the IoU metric. We detect, analyze and quantify an evaluation bias in the grounding of plural phrases and define a novel metric, c-IoU, based on a union box's component boxes. We experimentally show that our new metric greatly alleviates this bias and recommend using it for fairer evaluation of plural phrases in PG tasks.

show abstract

Section: Bias In Context Of System Performancementioning

confidence: 99%