corpus-id
stringlengths
15
17
image
imagewidth (px)
368
6.92k
pybv0228_p80.jpg
snbx0223_p21.jpg
zxfk0226_p12.jpg
txpp0227_p9.jpg
snbx0223_p43.jpg
ylvj0223_p20.jpg
xjpn0081_p3.jpg
snbx0223_p41.jpg
psyn0081_p29.jpg
snbx0223_p3.jpg
rnbx0223_p207.jpg
hldg0227_p6.jpg
hnhd0227_p7.jpg
jjvg0227_p5.jpg
flpp0227_p15.jpg
kqch0227_p2.jpg
snbx0223_p8.jpg
snbx0223_p14.jpg
snbx0223_p6.jpg
xjvw0217_p1.jpg
snbx0223_p9.jpg
hqgb0228_p0.jpg
fqvw0217_p33.jpg
jjvg0227_p7.jpg
snbx0223_p20.jpg
khxj0037_p3.jpg
txpp0227_p8.jpg
xkbv0228_p2.jpg
lycj0037_p7.jpg
grgb0228_p12.jpg
nhxj0037_p2.jpg
mxcj0037_p3.jpg
nrcj0037_p8.jpg
yshb0228_p20.jpg
hyhk0037_p4.jpg
lycj0037_p5.jpg
xpcf0004_p1.jpg
rzbj0037_p7.jpg
sncj0037_p2.jpg
nnhk0228_p0.jpg
hjxj0037_p1.jpg
phxj0037_p0.jpg
mybv0228_p12.jpg
ynbx0223_p14.jpg
hyhk0037_p5.jpg
khbw0217_p1.jpg
rnbx0223_p203.jpg
rzbj0037_p2.jpg
fgbd0079_p6.jpg
ynbx0223_p44.jpg
rgcw0217_p6.jpg
tnbx0223_p11.jpg
fygb0228_p0.jpg
tnbx0223_p129.jpg
sfmb0228_p78.jpg
mxxj0037_p1.jpg
trgj0223_p91.jpg
yxvw0217_p2.jpg
lkvw0217_p0.jpg
hnvw0217_p7.jpg
gsgj0223_p67.jpg
zzfv0228_p2.jpg
khxj0037_p0.jpg
txvw0217_p0.jpg
fxxj0037_p3.jpg
xqvw0217_p1.jpg
mtyj0226_p14.jpg
rgcw0217_p1.jpg
fqvw0217_p41.jpg
yhpj0226_p1.jpg
fqvw0217_p37.jpg
lkvw0217_p1.jpg
fqvw0217_p36.jpg
hhpj0226_p0.jpg
fyvw0217_p34.jpg
npvw0217_p5.jpg
kmwn0226_p17.jpg
zkww0217_p1.jpg
gfhd0082_p8.jpg
kfdw0217_p0.jpg
yrvw0217_p54.jpg
lgpj0226_p0.jpg
kfpj0226_p1.jpg
qjgn0226_p193.jpg
rycg0227_p6.jpg
hjpj0226_p2.jpg
nhpj0226_p0.jpg
mfpj0226_p0.jpg
gjpj0226_p5.jpg
ynbx0223_p67.jpg
mfnf0004_p0.jpg
xhfg0227_p40.jpg
lnbl0226_p8.jpg
npvw0217_p4.jpg
npvw0217_p3.jpg
mgbv0228_p12.jpg
fqvw0217_p38.jpg
sncw0217_p2.jpg
kgww0217_p14.jpg
ktdw0079_p0.jpg

Dataset Description

This is a VQA dataset based on Industrial Documents from MP-DocVQA dataset from MP-DocVQA.

Load the dataset

from datasets import load_dataset
import csv

def load_beir_qrels(qrels_file):
    qrels = {}
    with open(qrels_file) as f:
        tsvreader = csv.DictReader(f, delimiter="\t")
        for row in tsvreader:
            qid = row["query-id"]
            pid = row["corpus-id"]
            rel = int(row["score"])
            if qid in qrels:
                qrels[qid][pid] = rel
            else:
                qrels[qid] = {pid: rel}
    return qrels

corpus_ds = load_dataset("openbmb/VisRAG-Ret-Test-MP-DocVQA", name="corpus", split="train")
queries_ds = load_dataset("openbmb/VisRAG-Ret-Test-MP-DocVQA", name="queries", split="train")

qrels_path = "xxxx" # path to qrels file which can be found under qrels folder in the repo.
qrels = load_beir_qrels(qrels_path)
Downloads last month
143
Edit dataset card

Collection including openbmb/VisRAG-Ret-Test-MP-DocVQA