
컴퓨터 비전에서 카메라 이미지를 통해 Depth를 추정하는 것은 활발하게 연구되고 있다. 이때, Kitti에서 상위권에 차지하고 있는 논문 중 하나인 일명 BTS 논문에 대해서 리뷰해보려고 한다. Abstract 최근 single image(단안 이미지)에서 depth를 estimate 하는 문제를 CNN을 이용하면서 성능이 우수해지고 있다. 주로 encoder - decoder 구조를 이용한다. encoder : 기존 이미지에서 dense한 feature를 뽑는다. decoder : feature를 이용해 depth를 prediction encoder - decoder 구조를 효과적으로 사용하기 위해 다양한 기술들을 사용하며(skip connections, multi-layer deconvolution..

Semantic segmentation분야에서 높은 accuracy와 적은 inference time을 동시에 잡은 Network Abstract semantic segmentation 분야의 경우 low-level detail과 high-level semantics가 중요하다. 기존의 논문들은 inference 속도를 향상시키기 위해 low-level detail을 포기하게 된다. 본 논문에서는 spatial detail과 categorical semantics를 모두 충족시키는 네트워크를 제안하며 real-time으로 semantic segmentation을 진행한다. Detail branch에서는 넓은 channel과 얕은 layer를 이용해 high resolution을 얻고, Semantic b..
- Total
- Today
- Yesterday
- Camera
- 리뷰
- 딥러닝
- 컴퓨터비전
- slam
- 자율주행
- prediction
- 논문구현
- depth
- 연구
- mobile robot
- path planning
- 논문리뷰
- Vision
- segmentation
- Visual Odometry
- estimation
- 논문
- 아르떼뮤지엄 #여수 #여행
- bts
- 블로그 #작성요령 #시작
- opencv
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 |