Computer Vision with DL 입문

1. Basic computer vision 
   • • 02-07. 카메라의 구조00:53:46
     • Chapter 03. Image Processing04:24:39
     • 03-01. Local feature/Descriptor 검출00:50:45
     • 03-02. Local feature/Descriptor 검출 [실습]00:33:07
     • 03-03. Superpoint/Superglue 검출 [실습]00:18:21
     • 03-04. Global feature 검출00:32:35
     • 03-05. BoW 벡터 검출 [실습]00:37:00
     • 03-06. Learning based global feature 검출 [실습]00:43:34
     • 03-07. Feature tracking00:31:45
     • 03-08. Feature tracking [실습]00:17:32
     • Chapter 05. Multiple View Geometry12:35:51
     • 05-01. Epipolar geometry01:04:29
     • 05-02. Epipolar geometry [실습]00:16:38
     • 05-03. Homography00:26:54
     • 05-04. Homography [실습]00:26:54
     • 05-05. Simple monocular visual odometry [실습]00:57:17
     • 05-06. Triangulation00:40:03
     • 05-07. Triangulation [실습]00:23:21
     • 05-08. Perspective-n-Points (PnP), Direct Linear Transform (DLT)00:36:36
     • 05-09. Perspective-n-Points (PnP) [실습]00:26:40
     • 05-10. RANSAC00:56:58
     • 05-11. 최신 RANSAC00:39:32
     • 05-12. RANSAC [실습]00:28:02Basic computer vision
   • Basic Paper (참고)
     1. Gradient-based Learning Applied to Document Recognition (1998)
     2. Distinctive Image Features from Scale-Invariant Keypoints (2004)
     3. Histograms of Oriented Gradients for Human Detection (2005)
     4. SURF: Speeded Up Robust Features (2006)
2. Basic Deep learning https://cs231n.github.io/
3. Detection & Segmentation code 실행시켜보기
   • Yolov8
   • Yolact
   ⇒ 인터넷 자료에 무수한…. 한국어 트러블 슈팅 사용법들이 있으니 참고하면서 먼저 해보면 아~ 대충 이런 느낌이구나 하실 수 있슴다
4. (Must-무조건 읽으면 좋음 한국어 자료들과 같이 이해하면 빠르게 읽기 가능) Computer vision DL
   • Tip : CV DL 에서 중요한 점
     • CNN이 어떻게 탄생과, Transformer가 왜 등장했는지임
       • CNN 발전과정 중에서는 Objective function , Loss function 을 왜 바꿨고 어떤 Activate function을 사용했는지가 중요
       • Transformer 등장 후에는 또 어떤 식으로 vit → swin → deformable convolution → deformable detr → DeepViT 등으로 왜 발전했는지
   • Basic Classification
     1. ImageNet Classification with Deep Convolutional Neural Networks (2012)
     2. Very Deep Convolutional Networks for Large-Scale Image Recognition (2014)
     3. GoogLeNet – Going Deeper with Convolutions (2014)
     4. ResNet – Deep Residual Learning for Image Recognition (2015)
     5. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (2015)
     6. YOLO: You Only Look Once: Unified, Real-Time Object Detection (2016)
     7. Mask R-CNN (2017)
     8. EfficientNet – Rethinking Model Scaling for Convolutional Neural Networks (2019)
   • Vison transformer 계열 논문 : 최신 트랜드 (간단하게 말하면 이미지의 특징을 배경이든 작은 물체든 어떻게 잘 학습시킬 수 있는 구조)
   1.