TensorFlowObjectDetectionAPI-with-imgaug

Support Language

• Korean
• English

---

Release v1.2

Updated to last version of tf object detection API

INSTALLATION

1- Clone this repo

git clone https://github.com/JinLuckyboy/TensorFlowObjectDetectionAPI-with-imgaug.git

2- Clone tensorflow models repo

git clone https://github.com/tensorflow/models.git

3- Replace manually the files inside this folder repo into the Object Detection folder models/research/object_detection or run this file to do it automatically (both repos need to be in the same folder):

직접 이 저장소에 있는 파일을 Object Detection 폴더 models/research/object_detection로 덮어씌우거나, 이 파일을 실행해서 자동으로 하세요.(두 개의 클론한 저장소 폴더는 같은 폴더에 있어야 합니다.):

cd TensorFlowObjectDetectionAPI-with-imgaug
chmod +x patch.sh
./patch.sh

4- Install object detection API normally (creating the .protos first)

object detection API 일반 설치 (.proto 파일로부터 .py 파일을 생성합니다.)

cd ../models/research
protoc object_detection/protos/*.proto --python_out=.
cp object_detection/packages/tf2/setup.py .
python -m pip install --use-feature=2020-resolver .

# Verify installation was done correctly
python object_detection/builders/model_builder_tf2_test.py

You need to have imgaug installed pip install imgaug in order for everything to work.

If installation was sucessfull, you should get no errors and OK after finish

정상적으로 작동하기 위해서는 pip install imgaug로 imgaug를 설치해야 합니다.

설치가 성공했으면, Verify 단계에서 error가 없고 OK가 출력되어야 합니다.

HOW TO USE

---

Use augmentation

pipeline.config에 data_augmentation_options를 다음과 같이 설정하십시오.

Set data_augmentation_options in pipeline.config as follows (that applies for each model you are training)

train_config: {
  data_augmentation_options {
    random_imgaug {
      random_coef: 0.0
      # random_coef: [0, 1] 사이의 숫자를 가지며, 0이면 항상 적용하고, 1이면 항상 원본 이미지를 사용합니다. 이 옵션은 선택이므로 지워도 되며, 기본값은 0.0입니다.
      # random_coef: This option have a number between [0, 1]. If it is 0, augmented image is always used. If it is 1, original image is always used. It can be deleted. default: 0.0
    }
  }
}

---

Add augmentation options

object_detection/core/imgaug_utils.py를 열어 augmentation 옵션을 수정하십시오.

Open object_detection/core/imgaug_utils.py and Edit augmentation options.

Refer to imgaug documentation to get all possible augmentation options

augseq = iaa.Sequential([
    iaa.Crop(px=(0, 16)), # crop images from each side by 0 to 16px (randomly chosen)
    iaa.Fliplr(0.5), # horizontally flip 50% of the images
    iaa.GaussianBlur(sigma=(0, 3.0)) # blur images with a sigma of 0 to 3.0
])

REMEMBER!! Each time you change the augmentation options for imgaug_utils.py you MUST uninstall-reinstall object detection API for changes to take effect. You can do it yourself or use the script ./repack.sh for simplicity.

명심하세요!! 클론한 저장소에서 imgaug_utils.py를 수정하면 수정사항을 적용하기 위해 object detection API를 재설치해야 합니다. 직접 설치된 경로에서 수정하거나 ./repack.sh를 실행해서 간단하게 재설치할 수 있습니다.

---

Python versions that imgaug supports

The library uses python, which must be installed. Python 2.7, 3.4, 3.5, 3.6, 3.7 and 3.8 are supported.

imgaug가 지원하는 파이썬 버전: 2.7, 3.4, 3.5, 3.6, 3.7, 3.8

https://imgaug.readthedocs.io/en/latest/source/installation.html

---

참고가 될만한 사이트:

Recommanded Site:

• https://github.com/tensorflow/models/tree/master/research/object_detection
• https://github.com/aleju/imgaug
• https://imgaug.readthedocs.io/en/latest/source/overview_of_augmenters.html

---

2022년 01월 19일 github.com/tensorflow/models master브런치를 기준으로 제작되었습니다.

This is made based 2022-01-19 github.com/tensorflow/models master branch.