目录
- 1. 基础知识
- 2. 轨迹预测相关
- 2.1 Leapfrog Diffusion Model
- 2.2 MID
- 2.3 中科院
- 2.4 DICE
- 2.5 MotionDiffuser
- 2.6 DiffTraj
- 3. 练手时可参考
- 4. 扩展:扩散模型+RL
1. 基础知识
Stable diffusion扩散模型相关
Diffusion相关二
Diffusers是hugging face发起的用于专门用来收集扩散模型的模型库,就是包含了很多开源的扩散模型。
DiT - 基于Transformer架构的扩散模型
2. 轨迹预测相关
2.1 Leapfrog Diffusion Model
【CSDN博客】轨迹预测:Leapfrog Diffusion Model for Stochastic Trajectory Prediction
github
gitee
2.2 MID
【github项目链接】MID
we present a new framework to formulate the trajectory prediction task as a reverse process of motion indeterminacy diffusion (MID), in which we progressively discard indeterminacy from all the walkable areas until reaching the desired trajectory.
2.3 中科院
A multi-modal vehicle trajectory prediction framework via conditional diffusion model: A coarse-to-fine approach
2.4 DICE
【公众号文章】DICE:使用评分进行轨迹预测的多样化扩散模型
提出了一种利用扩散模型以计算高效的方式预测未来轨迹的新型框架,论文链接
2.5 MotionDiffuser
【论文链接】MotionDiffuser: Controllable Multi-Agent Motion Prediction using Diffusion
MotionDiffuser is a learned representation for the distribution of multi-agent trajectories based on diffusion models.
2.6 DiffTraj
【论文链接】DiffTraj: Generating GPS Trajectory with Diffusion Probabilistic Model
3. 练手时可参考
Diffusers代码级讲解(绪论)—— Diffusers快速上手
【初见Diffuser】利用PytorchLightning和HuggingFace训练你的第一个扩散模型
4. 扩展:扩散模型+RL
Diffusion Model + RL 系列技术科普博客(1):Diffuser
Diffusion Model + RL 系列技术科普博客(2):Decision Diffuser
Diffuser:Planning with Diffusion for Flexible Behavior Synthesis介绍
【论文翻译-RL×Diffusion】Planning with Diffusion for Flexible Behavior Synthesis
Sergey Levine 组的大作,中了 ICML 2022 年的 long talk。究竟是大佬整活,还是将扩散模型用于强化学习的开山之作呢?