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
- opencv
- rospackages
- ComputerVision
- gaussiansplatting
- catkinworkspace
- 3dmapping
- realtimerendering
- pointcloud
- LIDAR
- differentiablerendering
- vectorcalculus
- rosnoetic
- electromagnetism
- imageprocessing
- Slam
- tilebasedrasterizer
- adaptivedensitycontrol
- usbcamera
- 3dgaussiansplatting
- NERF
- covariancematrix
- raspberrypi
- sensorfusion
- rostopics
- turtlebot
- vectorfields
- turtlebot3
- alphablending
- roslaunch
- Ros
Archives
- Today
- Total
목록2D Vision/Segmentation Anything (1)
Wiredwisdom
Zero-Shot-Learning : 이미 충분히 학습이 된 네트워크(Foundation Model) 에서의 정보를 토대로 학습되지 못한 전혀 새로운 대상(혹은 의미)를 추론하여 학습된 모델을 통해 클래스로 분류하는 것을 의미한다.Few-Shot-Learning : 충분히 학습 된 Foundation Model에 전이 학습 등을 통하여 아주 소량의 라벨링 된 데이터틀 가지고 새로운 클래스를 만들어 Inference 시 인식하도록 하는 것이다. 이는 대량 학습을 위한 무수한 라벨링 데이터들에 대해 소모 되는 비용과 시간을 절약하기 위함이다. 기존의 분류체계만으로 새로운 학습에 적용하여 더 빠르게 학습하도록 유도하는 것이다. 이는 어느정도 인지 능력이 있는 사람의 빠른 학습 과정의 원리와 매우 유사하다. ..
2D Vision/Segmentation Anything
2024. 8. 2. 13:24