File size: 6,173 Bytes
6eece70 844a51a 6eece70 844a51a 6eece70 e30c694 6eece70 c1586cb 6eece70 fcc1085 c1586cb 6eece70 e30c694 6eece70 e30c694 6eece70 e30c694 6eece70 e30c694 6eece70 e30c694 c1586cb 6eece70 c1586cb 6eece70 c1586cb e30c694 c1586cb 6eece70 e30c694 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 |
import os
import warnings
from shutil import unpack_archive
from typing import Union, List
from urllib.request import urlretrieve
import pandas as pd
import sqlite3
import datasets
_CITATION = """@article{zanca2023contrastive,
title={Contrastive Language-Image Pretrained Models are Zero-Shot Human Scanpath Predictors},
author={Zanca, Dario and Zugarini, Andrea and Dietz, Simon and Altstidl, Thomas R and Ndjeuha, Mark A Turban and Schwinn, Leo and Eskofier, Bjoern},
journal={arXiv preprint arXiv:2305.12380},
year={2023}
}"""
_DESCRIPTION = """CapMIT1003 is a dataset of captions and click-contingent image explorations collected during captioning tasks.
CapMIT1003 is based on the same stimuli from the well-known MIT1003 benchmark, for which eye-tracking data
under free-viewing conditions is available, which offers a promising opportunity to concurrently study human attention under both tasks.
"""
_HOMEPAGE = "https://github.com/mad-lab-fau/CapMIT1003/"
MIT1003_URL = "http://people.csail.mit.edu/tjudd/WherePeopleLook/ALLSTIMULI.zip"
_VERSION = "1.0.0"
logger = datasets.logging.get_logger(__name__)
class CapMIT1003DB:
"""
Lightweight wrapper around CapMIT1003 SQLite3 database.
It provides utility functions for loading labeled images with captions and their associated click paths. To use it,
you first need to download the database from https://redacted.com/scanpath.db.
"""
def __init__(self, db_path: Union[str, bytes, os.PathLike] = 'capmit1003.db',
img_path: Union[str, bytes, os.PathLike] = os.path.join('mit1003', 'ALLSTIMULI')):
"""
Parameters
----------
db_path: str or bytes or os.PathLike
Path pointing to the location of the `scanpath.db` SQLite3 database.
img_path: str or bytes or os.PathLike
Path pointing to the location of the MIT1003 stimuli images.
"""
self.db_path = db_path
self.img_path = os.path.join(img_path, '')
if not os.path.exists(db_path) and not os.path.isfile(db_path):
warnings.warn('Could not find database at {}'.format(db_path))
if not os.path.exists(img_path) and not os.path.isdir(img_path):
warnings.warn('Could not find images at {}'.format(img_path))
def __enter__(self):
self.cnx = sqlite3.connect(self.db_path)
return self
def __exit__(self, type, value, traceback):
self.cnx.close()
def get_captions(self) -> pd.DataFrame:
""" Retrieve image-caption pairs of CapMIT1003 database.
Returns
-------
pd.DataFrame
Data frame with columns `obs_uid`, `usr_uid`, `start_time`, `caption`, `img_uid`, and `img_path`. See
accompanying readme for full documentation of columns.
"""
captions = pd.read_sql_query('SELECT * FROM captions o LEFT JOIN images i USING(img_uid)', self.cnx)
captions['img_path'] = self.img_path + captions['img_path']
return captions
def get_click_path(self, obs_uid: str) -> pd.DataFrame:
""" Retrieve click path for a specific image-caption pair.
Parameters
----------
obs_uid: str
The unique id of the image-caption pair for which to retrieve the click path.
Returns
-------
pd.DataFrame
Data frame with columns `click_id`, `obs_uid`, `x`, `y`, and `click_time`. See accompanying readme for full
documentation of columns.
"""
return pd.read_sql_query('SELECT x, y, click_time AS time FROM clicks WHERE obs_uid = ?', self.cnx,
params=[obs_uid])
class CapMIT1003(datasets.GeneratorBasedBuilder):
_URLS = [MIT1003_URL]
def _info(self):
return datasets.DatasetInfo(
description=_DESCRIPTION,
features=datasets.Features(
{
"obs_uid": datasets.Value("string"),
"usr_uid": datasets.Value("string"),
"caption": datasets.Value("string"),
"image": datasets.Image(),
"clicks_path": datasets.Sequence(datasets.Sequence(datasets.Value("int32"), length=2)),
"clicks_time": datasets.Sequence(datasets.Value("timestamp[s]"))
}
),
# No default supervised_keys (as we have to pass both question
# and context as input).
supervised_keys=None,
homepage=_HOMEPAGE,
citation=_CITATION,
)
def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
urls_to_download = {"mit1003": self._URLS}
downloaded_files = dl_manager.download_and_extract(urls_to_download)
downloaded_db = dl_manager.download({"cap1003": ["./capmit1003.db"]})
return [
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"mit1003_path": downloaded_files["mit1003"], "capmit1003_db_path": downloaded_db["cap1003"]}),
]
def _generate_examples(self, mit1003_path, capmit1003_db_path):
with CapMIT1003DB(os.path.join(capmit1003_db_path[0]), os.path.join(mit1003_path[0], "ALLSTIMULI")) as db:
image_captions = db.get_captions()
for pair in image_captions.itertuples(index=False):
obs_uid = pair.obs_uid
click_path = db.get_click_path(obs_uid)
xy_coordinates = click_path[['x', 'y']].values
clicks_time = click_path["time"].values
example = {
"obs_uid": obs_uid,
"usr_uid": pair.usr_uid,
"image": pair.img_path,
"caption": pair.caption,
"clicks_path": xy_coordinates,
"clicks_time": clicks_time
}
yield obs_uid, example
#datasets-cli test CapMIT1003/CAPMIT1003.py --save_info --all_configs |