Notice
Recent Posts
Recent Comments
Link
| 일 | 월 | 화 | 수 | 목 | 금 | 토 |
|---|---|---|---|---|---|---|
| 1 | ||||||
| 2 | 3 | 4 | 5 | 6 | 7 | 8 |
| 9 | 10 | 11 | 12 | 13 | 14 | 15 |
| 16 | 17 | 18 | 19 | 20 | 21 | 22 |
| 23 | 24 | 25 | 26 | 27 | 28 | 29 |
| 30 |
Tags
- Mask-and-replace diffusion strategy
- Discrete diffusion
- img2pose: Face Alignment and Detection via 6DoF
- VQ-diffusion
- ENERGY-BASED MODELS FOR CONTINUAL LEARNING
- Mask diffusion
- CIL
- Face Alignment
- Face Pose Estimation
- learning to prompt for continual learning
- state_dict()
- Facial Landmark Localization
- Class Incremental Learning
- Energy-based model
- learning to prompt
- CVPR2022
- L2P
- Continual Learning
- 베이지안 정리
- mmcv
- PnP algorithm
- Img2pose
- VQ-VAE
- Markov transition matrix
- prompt learning
- DualPrompt
- timm
- requires_grad
- Class Incremental
- Vector Quantized Diffusion Model for Text-to-Image Synthesis
Archives
- Today
- Total
Computer Vision , AI
[One-page summary] MimicPlay: Long Horizon Imitation Learning by Watching Human Play by Wang et al. 본문
Paper_review[short]
[One-page summary] MimicPlay: Long Horizon Imitation Learning by Watching Human Play by Wang et al.
Elune001 2024. 1. 16. 00:41● Summary: Plan from human play, Control from teleoperated demonstrations
● Approach highlight
-
high-level planer and low-level control policy: learn high-level plans from human play data with their hands and learn


-
low-level control using a small number of teleoperated demonstrations

● Main Results

● Discussion
- Efficiency in sophisticated finger-driven tasks (high-level planner can work?)
- What is high-level? (ambiguous definitions)