MultiviewC

MultiviewC 数据集主要用于多视图牛动作识别、3D 物体检测和跟踪。我们基于 CISRO 提供的真实牛视频数据集，通过 UE4 构建了一个新颖的合成数据集 MultiviewC。我们的数据集的格式已经在 MultiviewX 的基础上进行了调整，用于设置、注释和文件结构。
MultiviewC 数据集是在 37.5 米 x 37.5 平方米的场地上生成的。它包含 7 个监控牛活动的摄像机。 MultiviewC 中的图像具有 1280x720 的高分辨率，我们数据集中的合成动物非常逼真。

The MultiviewC dataset is primarily used for multi-view cattle action recognition, 3D object detection and tracking. We constructed a novel synthetic dataset named MultiviewC based on the real-world cattle video dataset provided by CISRO using UE4. The format of our dataset has been adjusted based on MultiviewX for setting, annotation and file structure.

The MultiviewC dataset is generated on a 37.5 m × 37.5 m square field. It comprises 7 cameras monitoring cattle activities. The images in MultiviewC have a high resolution of 1280×720, and the synthetic animals in our dataset are highly realistic.

搜集汇总

背景与挑战

背景概述

MultiviewC是一个用于多视图牛动作识别、3D物体检测和跟踪的合成数据集，基于真实牛视频数据构建，包含7个摄像机拍摄的高分辨率图像，覆盖大面积场地，合成动物逼真。