Facemap: a framework for modeling neural activity based on orofacial tracking

Neural activity, face camera data, and behavioral tracking from 16 large-scale recordings from visual cortex, and the train/test data for the Facemap keypoint tracker. Please see the paper and code repository for more details about the data acquisition and analysis.<br>Data files:neural_activity_*.zip: contains npz files with deconvolved neural traces "spks" (neurons by times), neuron positions "xpos" and "ypos", camera timestamps "tcam", neural timestamps "tneural", and running speed "run"filtered_keypoints.zip: contains npy files with a dictionary with filtered keypoint traces "xy" and keypoint labels "keypoint_labels", which can be used to predict neural activitysvds.zip: contains npy files which are dictionaries in Facemap output format for the motion and movie SVD of the camera recordingcam.zip: contains camera recordings in mp4 and avi format, and the raw keypoint tracking h5 filespose_estimation.zip: dataset used for training and testing the Facemap keypoint tracking network. The zipped file contains train and test files: mouse face images in PNG format and h5 annotations with "x" coordinates and "y" coordinates for the 15 labeled keypoints. Each subfolder in train and test is formatted as either ‘mousename_date_cameraview' or ‘cameraview_mousename_date'.If you use this data please cite this data repository and the original paper:<br>Syeda, A., Zhong, L., Tung, R., Long, W., Pachitariu, M., &amp; Stringer, C. (2023). Facemap: a framework for modeling neural activity based on orofacial tracking. <i>Nature Neuroscience</i>.<br>

本数据集涵盖来自视觉皮层的16次大规模神经记录、面部摄像头数据与行为追踪数据，以及Facemap关键点跟踪器的训练与测试数据。如需了解数据采集与分析的详细信息，请参阅对应论文与代码仓库。

各数据文件说明如下：
- `neural_activity_*.zip`：内含npz格式的解卷积神经活动轨迹文件，包含神经元 spike 轨迹"spks"（神经元×时间维度矩阵）、神经元空间位置"xpos"与"ypos"、摄像头时间戳"tcam"、神经时间戳"tneural"以及运行速度"run"。
- `filtered_keypoints.zip`：内含npy格式文件，其内部为存储过滤后关键点轨迹"xy"与关键点标签"keypoint_labels"的字典，可用于神经活动预测。
- `svds.zip`：内含npy格式文件，为符合Facemap输出格式的摄像头记录运动与影片奇异值分解（SVD）字典。
- `cam.zip`：内含mp4与avi格式的摄像头录制文件，以及原始关键点跟踪h5文件。
- `pose_estimation.zip`：用于训练与测试Facemap关键点跟踪网络的数据集。该压缩包包含训练集与测试集文件：PNG格式的小鼠面部图像，以及包含15个标注关键点x、y坐标的h5注释文件。训练集与测试集下的每个子文件夹均采用"小鼠名_日期_摄像头视角"或"摄像头视角_小鼠名_日期"的命名格式。

若使用本数据集，请引用本数据仓库与原始论文：
Syeda, A., Zhong, L., Tung, R., Long, W., Pachitariu, M., & Stringer, C. (2023). 《Facemap：基于口面部追踪的神经活动建模框架》，《自然神经科学》（*Nature Neuroscience*）。