corpus-id
stringlengths
5
9
image
imagewidth (px)
650
1.4k
6519.png
1464.png
6074.png
30972.png
5137.png
1851.png
1738.png
5629.png
11916.png
12611.png
31759.png
10923.png
24682.png
12005.png
26174.png
384.png
14847.png
11750.png
5581.png
22682.png
2516.png
30518.png
15960.png
10593.png
2652.png
5998.png
22692.png
21662.png
9104.png
20877.png
11461.png
14305.png
9530.png
13211.png
26831.png
3219.png
1869.png
23757.png
14344.png
3788.png
11535.png
32219.png
17482.png
11617.png
23477.png
26658.png
8111.png
7948.png
13301.png
14095.png
33570.png
25130.png
33488.png
21444.png
7882.png
28403.png
2725.png
6435.png
8611.png
8969.png
3640.png
22768.png
678.png
6681.png
14298.png
19548.png
6531.png
15985.png
8876.png
26399.png
191.png
14127.png
28751.png
9909.png
29102.png
6271.png
29101.png
10927.png
22360.png
9048.png
30674.png
28156.png
1148.png
21519.png
3329.png
32233.png
4506.png
5038.png
28024.png
3956.png
30343.png
7249.png
7381.png
9228.png
14210.png
30094.png
24864.png
11576.png
30670.png
11038.png

Dataset Description

This is a VQA dataset based on Scientific Plots from PlotQA dataset from PlotQA.

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-PlotQA", name="corpus", split="train")
queries_ds = load_dataset("openbmb/VisRAG-Ret-Test-PlotQA", 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
128

Collection including openbmb/VisRAG-Ret-Test-PlotQA