clane9/NSD-Flat
收藏资源简介:
NSD-Flat是一个Hugging Face数据集,包含从Natural Scenes Dataset中预处理的大脑活动平面图,这些图被限制在视觉皮层感兴趣区域,并以PNG图像的形式呈现。数据集的特征包括subject_id、trial_id、session_id、nsd_id、image、activity、subject、flagged、BOLD5000、shared1000、coco_split、coco_id、objects、captions和repetitions等。数据集分为训练集和测试集,分别包含195,000和18,000个样本。数据集的构建过程包括下载源数据、转换COCO注释和生成数据集。
NSD-Flat is a Hugging Face dataset that provides preprocessed flat maps of brain activity extracted from the Natural Scenes Dataset. These maps are restricted to the regions of interest (ROIs) of the visual cortex and are presented in PNG image format. The dataset includes the following features: subject_id, trial_id, session_id, nsd_id, image, activity, subject, flagged, BOLD5000, shared1000, coco_split, coco_id, objects, captions, and repetitions. It is divided into training and test subsets, which contain 195,000 and 18,000 samples respectively. The construction process of the dataset includes downloading the source data, converting COCO annotations, and generating the dataset.
数据集概述
数据集名称
- NSD-Flat
数据集特征
- 基本特征:
subject_id: int64trial_id: int64session_id: int64nsd_id: int64image: imageactivity: imagesubject: stringflagged: boolBOLD5000: boolshared1000: boolcoco_split: stringcoco_id: int64
- 对象结构:
objects: structarea: sequence: int64bbox: sequence: sequence: float64category: sequence: stringiscrowd: sequence: int64segmentation: listcounts: stringpoly: sequence: sequence: float64size: sequence: int64
supercategory: sequence: stringtarget: sequence: int64
- 重复结构:
repetitions: structsubject1_rep0tosubject8_rep2: int64
- 描述:
- 包含预处理的脑活动平面图,源自Natural Scenes Dataset,限定于视觉皮层感兴趣区域,并以PNG图像形式呈现。
数据集划分
- 训练集:
- 大小: 26695182666.0 bytes
- 样本数: 195000
- 测试集:
- 大小: 2461280671.0 bytes
- 样本数: 18000
下载与数据集大小
- 下载大小: 22565691383 bytes
- 数据集大小: 29156463337.0 bytes
任务类别
- image-to-image
- object-detection
标签
- biology
- neuroscience
- fmri
大小类别
- 100K<n<1M
许可证
引用信息
-
Allen et al. (2022):
@article{allen2022massive, title = {A massive 7T fMRI dataset to bridge cognitive neuroscience and artificial intelligence}, author = {Allen, Emily J and others}, journal = {Nature neuroscience}, volume = {25}, number = {1}, pages = {116--126}, year = {2022}, publisher = {Nature Publishing Group US New York} }
-
Lane (2023):
@misc{lane2023nsdflat, author = {Connor Lane}, title = {NSD-Flat: Pre-processed brain activity flat maps from the Natural Scenes Dataset}, howpublished = {url{https://huggingface.co/datasets/clane9/NSD-Flat}}, year = {2023}, }




