This repository contains the official data for the paper "VQPP: Video Query Performance Prediction Benchmark".
VQPP is a benchmark designed to evaluate methods that estimate the difficulty of a text query for video retrieval systems. It standardizes the evaluation of Pre-retrieval and Post-retrieval QPP estimators across two datasets (MSR-VTT, VATEX) and two state-of-the-art retrieval architectures (GRAM, VAST).… See the full description on the dataset page:
https://huggingface.co/datasets/funzon3/VQPP.
