-
기간 : 22.07.04 ~ 08.15 (약 2개월, 이후 기업프로젝트 진행)
-
각 프로젝트들은
[EX- no.]-[E-no.] Title의 형식으로 구성되어있고, 해당 Folder내에 관련 자료들이 있습니다.
| No. | Title | Concept | Description | Link |
|---|---|---|---|---|
| 1 | Implementation of ResNet Papers | ResNet | ResNet-34, ResNet-50 논문 구현, ResNet 모델을 활용하여 Image Classification 모델 훈련 | Link |
| 2 | Compare Augmentation | Augmentation | 기본 Augment과 CutMix, MixUp, Cutout 기법적용 후 비교 (+ Mixup & Cutmix, cutout + labelsmoothing 적용 Test) |
Link |
| 3 | Make Object Detection System(autonomous driving) | Object Detection | Object Detection 모델을 이용한 자율주행 보조장치(일정조건시 STOP)시스템 구현 | Link |
| 4 | Segmentation(U-Net, U-Net++) | Segmentation, U-Net | U-Net, U-Net++ 논문 구현, KITTI 데이터셋 Segmentation 진행 | Link |
| 5 | Make CAM & Grad-CAM | CAM, Grad-CAM | CAM, Grad-CAM을 구현하고 이미지별 모델 참고부분 확인 | Link |
| 6 | make ocr | OCR | keras-ocr detector와 CRNN recognizer를 엮어 원본 이미지 입력으로부터 text가 출력되는 OCR 생 | Link |
| 7 | make action sticker | Human Pose Estimation | TFRecord 파일을 Hourglass 모델에 넣어 학습 후 포즈 확인 | Link |