Underwater images collected by an Autonomous Surface Vehicle in Grandfond, Réunion - 2023-02-03

This dataset was collected by an Autonomous Surface Vehicle in Grandfond, Réunion - 2023-02-03.

Underwater or aerial images collected by scientists or citizens can have a wide variety of use for science, management, or conservation.
These images can be annotated and shared to train IA models which can in turn predict the objects on the images.
We provide a set of tools (hardware and software) to collect marine data, predict species or habitat, and provide maps.

Image acquisition
This session has 16.92 GB of MP4 files, which were trimmed into 8221 frames (at 2997/1000 fps).
The frames are georeferenced.
98.1% of these extracted images are useful and 1.9% are useless, according to predictions made by Jacques model.
Multilabel predictions have been made on useful frames using DinoVd'eau model.

GPS information:
The data was processed with a PPK workflow to achieve centimeter-level GPS accuracy.
Base : Files coming from rtk a GPS-fixed station or any static positioning instrument which can provide with correction frames.
Device GPS : Emlid Reach M2
Quality of our data - Q1: 59.69 %, Q2: 4.3 %, Q5: 36.01 %

Bathymetry

The data are collected using a single-beam echosounder ETC 400.

We only keep the values which have a GPS correction in Q1.
We keep the points that are the waypoints.

We keep the raw data where depth was estimated between 0.2 m and 50.0 m deep.
The data are first referenced against the WGS84 ellipsoid. Then we apply the local geoid if available.
At the end of processing, the data are projected into a homogeneous grid to create a raster and a shapefiles.
The size of the grid cells is 0.133 m.
The raster and shapefiles are generated by linear interpolation. The 3D reconstruction algorithm is ballpivot.

Generic folder structure

YYYYMMDD_COUNTRYCODE-optionalplace_device_session-number
├── DCIM : folder to store videos and photos depending on the media collected.
├── GPS : folder to store any positioning related file. If any kind of correction is possible on files (e.g. Post-Processed Kinematic thanks to rinex data) then the distinction between device data and base data is made. If, on the other hand, only device position data are present and the files cannot be corrected by post-processing techniques (e.g. gpx files), then the distinction between base and device is not made and the files are placed directly at the root of the GPS folder.
│ ├── BASE : files coming from rtk station or any static positioning instrument.
│ └── DEVICE : files coming from the device.
├── METADATA : folder with general information files about the session.
├── PROCESSED_DATA : contain all the folders needed to store the results of the data processing of the current session.
│ ├── BATHY : output folder for bathymetry raw data extracted from mission logs.
│ ├── FRAMES : output folder for georeferenced frames extracted from DCIM videos.
│ ├── IA : destination folder for image recognition predictions.
│ └── PHOTOGRAMMETRY : destination folder for reconstructed models in photogrammetry.
└── SENSORS : folder to store files coming from other sources (bathymetry data from the echosounder, log file from the autopilot, mission plan etc.).

Software
All the raw data was processed using our worflow. All predictions were generated by our inference pipeline. You can find all the necessary scripts to download this data in this repository. Enjoy your data with SeatizenDOI!

本数据集于2023年2月3日在留尼旺岛Grandfond海域由自主水面航行器（Autonomous Surface Vehicle）采集。
科研人员或公民采集的水下、航空影像可广泛应用于科学研究、资源管理与生态保护领域。此类影像经标注后可用于训练人工智能（AI）模型，以实现影像内目标的自动识别。我们提供一套涵盖硬件与软件的工具集，用于海洋数据采集、物种/栖息地识别以及地图生成。

图像采集
本次采集任务生成了总容量16.92GB的MP4视频文件，经剪辑后提取得到8221帧影像（帧率为2997/1000 fps）。所有提取帧均已完成地理配准。根据Jacques模型的预测结果，其中98.1%的影像为有效帧，1.9%为无效帧。针对有效帧，我们采用DinoVd'eau模型完成了多标签分类预测。

GPS定位信息
本数据集采用后处理动态差分（Post-Processed Kinematic, PPK）处理流程，以实现厘米级的GPS定位精度。
- 基准站数据：来自RTK（Real-Time Kinematic）固定GPS站或可提供校正帧的静态定位设备的文件。
- 设备GPS：采用Emlid Reach M2型号设备。
本数据集的质量分布为：Q1占比59.69%，Q2占比4.3%，Q5占比36.01%。

测深数据
本次测深数据采用单波束测深仪（single-beam echosounder）ETC 400采集。
我们仅保留了Q1等级且带有GPS校正信息的测点数据，同时保留所有航迹点数据。此外，我们仅保留水深估算值介于0.2米至50.0米之间的原始测深数据。
数据处理流程为先以WGS84椭球体（WGS84 ellipsoid）为基准进行坐标转换，若存在本地大地水准面则进一步应用其进行高程修正。最终将所有数据投影至统一格网，生成栅格（raster）文件与形状文件（shapefiles），格网单元尺寸为0.133米。上述栅格与形状文件通过线性插值生成，三维重建算法采用球枢算法（ballpivot）。

标准文件夹组织结构
本数据集采用如下命名格式：YYYYMMDD_COUNTRYCODE-optionalplace_device_session-number
其目录结构如下：
├── DCIM：用于存储采集到的视频与照片文件的目录。
├── GPS：用于存储所有定位相关文件的目录。若可对文件进行差分校正（例如基于RINEX数据的后处理动态差分），则需将基准站数据与设备数据分开存储；若仅存在设备位置数据且无法通过后处理技术进行校正（例如GPX文件），则无需区分基准站与设备数据，直接将所有文件置于GPS目录根目录下。
│ ├── BASE：存储来自RTK基准站或静态定位设备的基准站数据。
│ └── DEVICE：存储来自采集设备的设备定位数据。
├── METADATA：存储本次采集任务通用信息文件的目录。
├── PROCESSED_DATA：存储本次采集任务数据处理结果的目录，包含以下子目录：
│ ├── BATHY：存储从任务日志中提取的原始测深数据的输出目录。
│ ├── FRAMES：存储从DCIM目录视频中提取的地理配准帧影像的输出目录。
│ ├── IA：存储图像识别模型预测结果的目标目录。
│ └── PHOTOGRAMMETRY：存储摄影测量重建模型的目标目录。
└── SENSORS：存储其他来源传感器数据的目录，例如测深仪采集的测深数据、自动驾驶仪日志文件、任务规划文件等。

软件处理说明
所有原始数据均通过我们的标准化工作流进行处理，所有模型预测结果均由我们的推理管线生成。你可在本代码仓库中获取下载该数据集所需的全部脚本。欢迎使用SeatizenDOI进行数据引用！