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 |
- 논문분석
- object detection
- TensorFlow Object Detection Error
- VOC 변환
- Towards Deep Learning Models Resistant to Adversarial Attacks
- DACON
- Git
- 객체 탐지
- paper review
- 사회초년생 추천독서
- 리눅스 빌드
- 개발흐름
- TensorFlow Object Detection API install
- TensorFlow Object Detection Model Build
- 크롤링
- 커스텀 애니메이션 적용
- 기능과 역할
- TensorFlow Object Detection 사용예시
- Docker
- DOTA dataset
- Branch 활용 개발
- AI Security
- Linux build
- 논문 분석
- Paper Analysis
- Carla
- Object Detection Dataset 생성
- InstructPix2Pix
- Custom Animation
- CARLA simulator
- Today
- Total
목록Deap learning (10)
JSP's Deep learning
일반적인 설치 방법으로 설치되지 않을 때, 다음의 명령어를 통해서 정상적인 설치가 가능하다. (기존에 설치가 된 경우에도 사용가능하다) # 1. pip3 install --upgrade --force-reinstall --no-cache-dir jupyter # 2. pip3 uninstall notebook # 3. pip3 install --upgrade notebook

mmdetection이란? Tensorflow object detection API처럼 다양한 객체 탐지 모델을 제공하고 학습, 추론, 파인튜닝, 모델 수정 등 다양한 기능을 제공하는 open source이며, pytorch을 기반으로 구성되어있다. 1. mmdetection 다운로드 git clone https://github.com/open-mmlab/mmdetection.git 2. Dockerfile 빌드 GPU 이름 확인 nvidia-smi --query | fgrep 'Product Name' CUDA GPU와 호환되는 Compute Capability 확인(링크) GTX3090 : 8.6 pytorch docker image을 통한 버전확인 pytorch : 1.9.0 cuda : 11.1 ..
객체 탐지 모델의 성능은 mAP(mean Average Precision) 값을 통해서 평가할 수 있다. 본문에서는 mAP을 구하는 함수를 직접 구현한다. Ground-Truth Bounding Box 하나는 (xmin, ymin, xmax, ymax, label)로 구성되고, Prediction Bounding Box 하나는 (xmin, ymin, xmax, ymax, confidence score, label)로 구성되는 것을 전제로 코드를 작성한다. mAP의 함수에 입력되는 GT-BBOX, Pred-BBOX는 모든 이미지에 대한 BBOX 정보를 포함한다. 1. 패키지 로드 import numpy as np 2. 함수 정의 2.1. Label 획득 함수 모델이 추론할 수 있는 Label의 경우에 사전..

1. 'cv2' import Error AttributeError: partially initialized module 'cv2' has no attribute '_registerMatType' (most likely due to a circular import) cv2 패키지 폴더 삭제 후 다음의 명령어 실행 (패키지 재설치) 2. Matplotlib Error Jupyter notebook의 ipynb 파일에서 plt.show()에도 이미지가 표기되지 않는 Error Matplotlib 패키지 import 전에 %matplotlib inline 코드 추가 3. Permission Error Read-only file error로 mkdir 명령어가 거부되는 경우(Mac OS) Mac을 재부팅한다. 재..

데이터 생성부터 평가까지 모든 과정 1. 커스텀 데이터 셋 생성 이미지 다운로드 labelImg 실행 cd TensorFlow/addons/labelImg python labelImg.py 이미지 로드 및 annotations annotated 이미지 저장 데이터 셋 분할 cd TensorFlow/scripts/preprocessing python partition_dataset.py –x –i Users/jiseung/Tensorflow/workspace/training_demo/images –r 0.1 label map 생성 TensorFlow Records 생성 2. 모델 학습 모델 다운로드 및 저장(링크 : Download) pipeline 파일 복사 및 저장 pipeline.config 수정 스..

1. Pre-Trained Model 다운로드 다운로드 링크 : Download TensorFlow/workspace/training_demo/pre-trained-models에 압축해제 압축해제시 폴더 구조 2. Training pipeline 파일 생성 Tensorflow/workspace/training_demo/models에 폴더 이름을 지정(사용자정의)하여 생성 생성한 폴더에 pre-trained model의 pipeline.config 파일 저장 폴더 구조 예시 3. pipeline.config 파일 조정 Model이 가지는 classes 수 설정(즉, 모델이 분류할 수 있는 클래스 수) Train batch_size 조정(그래픽 카드 성능에 맞게 조정) Pre-trained model che..

1. Detection 데이터 만들기 1) labelImg를 이용한 이미지 라벨링 진행순서 labelImg 실행 TensorFlow/addons/labelImg 경로로 이동 python labelImg.py [이미지 데이터 경로] (본문에서는 TensorFlow/workspace/training_demo/images) 이미지 파일 선택 Ctrl + 마우스 클릭으로 bounding box 지정 Label 입력 or Default label 지정 저장(본문에서는 TensorFlow/workspace/training_demo/images에 저장) -> 최종적으로 Image에 대한 box 및 label 정보가 .xml 형태로 저장 labelImg 예시 실행 결과 및 바운딩 박스 표기 labelImg로 생성된 ...

TensorFlow Object Detection API를 사용하기 위해서 아나콘다 환경을 구성한다. 아나콘다 환경이 필수적인 것은 아니지만 TensorFlow의 버전을 관리하기에 용이하므로 아나콘다 환경을 사용한다. 1. Anaconda 환경 설치 다운로드 및 설치 다운로드 링크 : https://www.anaconda.com/products/distribution#macos Anaconda Prompt 실행(Mac에서는 설치 후 Mac 터미널을 실행하면 곧바로 적용된다) 가상환경 생성 (TensorFlow Object Detection API에서는 파이썬 버전 3.7이상을 권장한다) conda create –-name “가상환경명” python=“version” 가상환경 생성 확인 conda env ..