Category
EasyCraft
Paper

EasyCraft

解决的目标: 自动化角色定制,从不同模态的信息输入中生成不同引擎风格下的3d角色 存在的问题: 输入模态,风格的多样性:...
Avatar photo
Weakly Supervised Video Anomaly Detection and Localization with Spatio-Temporal Prompts
Paper

Weakly Supervised Video Anomaly Detection and Localization with Spatio-Temporal Prompts

目的: 时间和空间维度的分离处理 方法: 时间异常和空间异常,蓝线时间,绿线空间 时间异常:双分支结构,基于CLIP,分...
Avatar photo
Video Anomaly Detection in 10 Years: A Survey and Outlook
Paper

Video Anomaly Detection in 10 Years: A Survey and Outlook

Video Anomaly Detection in 10 Years: A Survey and Outlook 问题...
Avatar photo
Diff3DETR: Agent-based Diffusion Model for Semi-supervised 3D Object Detection
Paper

Diff3DETR: Agent-based Diffusion Model for Semi-supervised 3D Object Detection

目的: 3D目标检测,通过教师-学生模型为未标注的点云生成伪标签,缓解数据集不足的问题 方法: 采用教师学生模型来解决标...
Avatar photo
LORA: LOW-RANK ADAPTATION OF LARGE LANGUAGE MODELS
Paper

LORA: LOW-RANK ADAPTATION OF LARGE LANGUAGE MODELS

目的: 预训练模型的微调伴随着模型参数的增大带来了更大的压力,本文提出了lora,冻结了预训练的权重,通过在transf...
Avatar photo
OpenGaussian: Towards Point-Level 3D Gaussian-based Open Vocabulary Understanding
Paper

OpenGaussian: Towards Point-Level 3D Gaussian-based Open Vocabulary Understanding

目的: 3D点级别的开放词汇理解,将3D点链接到2D掩码 方法: 3D一致性的实例特征学习: 与之前的 方法类似,为每一...
Avatar photo
LangSplat: 3D Language Gaussian Splatting
Paper

LangSplat: 3D Language Gaussian Splatting

目的: 建模一个三维语言场,使用户能够使用开放性语言与三维世界进行交互。用3DGS中的高斯去编码从CLIP中提取的语言特...
Avatar photo
LERF: Language Embedded Radiance Fields
Paper

LERF: Language Embedded Radiance Fields

目的: 将语言嵌入辐射场,明确的说,将CLIP嵌入到Nerf中,支持在3D空间中进行开放式语言查询 方法: 通过沿训练光...
Avatar photo
Segment Any 3D Gaussians
Paper

Segment Any 3D Gaussians

目的: 利用2D视觉提示对3D目标进行分割,通过对每一个3D高斯点附加一个尺度门控亲和特征来实现 方法: 为每个高斯点设...
Avatar photo
Segment Anything in 3D with NeRFs
Paper

Segment Anything in 3D with NeRFs

目的: 采用SAM生成的2D掩膜,借助Nerf生成物体的3D掩膜 方法: 前置知识: Nerf: 利用一个多视角的2D图...
Avatar photo