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 | 31 |
Tags
- NERF
- vectorcalculus
- tilebasedrasterizer
- turtlebot
- catkinworkspace
- rostopics
- sensorfusion
- opencv
- LIDAR
- ComputerVision
- electromagnetism
- realtimerendering
- 3dgaussiansplatting
- pointcloud
- Slam
- differentiablerendering
- 3dmapping
- roslaunch
- raspberrypi
- Ros
- rospackages
- rosnoetic
- covariancematrix
- vectorfields
- usbcamera
- imageprocessing
- adaptivedensitycontrol
- alphablending
- gaussiansplatting
- turtlebot3
Archives
- Today
- Total
Wiredwisdom
Depthwise Convolution 본문

본래의 컨볼루션은 커널이 입력측의 채널을 전부 컨볼루션을 하고 더하는 방식으로 이루어지나
MobileNet-V2 에서는 각각의 입력 채널에 맞는 커널이 각기 따로 계산이 되며 합산은 Pointwise convolution 을 통해 이루어지게 된다.
이 두가지 주요 과정을 통하여 Operation이 줄어들게 된다.
임베딩 시스템에서는 이런 방식의 네트워크를 주로 사용한다.
이에 대한 내용은 차후 MobileNet-V2에서 자세히 다루도록한다.
'2D Vision > CNN (Object Detection)' 카테고리의 다른 글
| Layer Normalization (0) | 2025.06.27 |
|---|---|
| Alexnet Architecture (2012) (0) | 2025.06.27 |
| LeNet-5 (1998) (0) | 2025.03.24 |