Surrey: AI to help turn dog pics into 3D models
萨里：人工智能将狗狗照片转化为 3D 模型
Photographs of dogs could be used to help generate 3D models more accurately, a study has revealed.
一项研究表明，狗的照片可以用来帮助更准确地生成 3D 模型。
Researchers at the University of Surrey taught an artificial intelligence (AI) system to predict the 3D pose from a 2D image of a dog.
萨里大学的研究人员训练了一种人工智能（AI）系统，可以从狗的 2D 图像中预测 3D 姿势。
A myriad of virtual dogs were created using the video game Grand Theft Auto.
电子游戏侠盗猎车手创造了无数的虚拟狗。
Postgraduate research student Moira Shooter said: “From ecology to animation, this neat solution has so many possible uses.”
学术型研究生 Moira Shooter 说：“从生态学到动画，这种巧妙的解决方案有很多理论上可能性的用途。”
One way to teach AI to get 3D information from 2D images is to show it photos while giving it information about 3D ‘ground truth’ – where the objects are in 3D space.
训练人工智能从 2D 图像中获取 3D 信息的一种方法是向其展示照片，同时向其提供有关 3D “地面实况”的信息——物体在 3D 空间中的位置。
For humans, that means wearing motion capture suits.
对人类来说，这意味着要穿动作捕捉服。
Unable to replicate the same with dogs, researchers altered the code of Grand Theft Auto V, switching the main character for one of eight breeds of dog – a process known as modding.
由于无法在狗身上复制同样的东西，研究人员修改了《侠盗猎车手V》的代码，将主角换成了八种狗中的一种，这一过程被称为改装。
They generated 118 videos of the dogs sitting, walking, barking and running in a range of different weather and lighting conditions.
他们制作了 118 段狗在不同天气和照明条件下坐着、走路、吠叫和奔跑的视频。
The team called their new database DigiDogs, which was made up of 27,900 frames.
该团队称他们的新数据库为 DigiDogs ，该数据库由 27900 帧组成。
They now plan to fine tune the system using Meta’s DINOv2 model to make sure it can predict a 3D pose just as well from real dog pictures.
他们现在计划使用 Meta 的 DINOv2 模型对系统进行微调，以确保它能够从真实的狗照片中预测出 3D 姿势。
“Our model was trained on CGI dogs, but we were able to use it to make 3D skeletal models from photographs of real animals,” Ms Shooter said.
Shooter 女士说：“虽然我们的模型是在 CGI 狗身上训练的，但我们能够用它从真实动物的照片中制作 3D 骨骼模型。”
“That could let conservationists spot injured wildlife, or help artists create more realistic animals in the metaverse.
“这可以让自然保护主义者发现受伤的野生动物，或者帮助艺术家在元宇宙中创造更逼真的动物。
She added: “3D poses contain so much more information than 2D photographs.”
她补充道：“ 3D 姿势包含的信息比 2D 照片多得多。”
原文：https://www.bbc.com/news/uk-england-surrey-68435966
仅供学习参考
