시맨틱 세분화
Roboflow에 호스팅된 시맨틱 세분화 모델에서 추론을 실행하세요.
Linux 또는 MacOS
로컬 파일 YOUR_IMAGE.jpg
:
base64 YOUR_IMAGE.jpg | curl -d @- \
"https://segment.roboflow.com/your-model/42?api_key=YOUR_KEY"
웹에 호스팅된 이미지의 URL을 통해 추론하기 ( URL 인코딩을 잊지 마세요):
curl -X POST "https://segment.roboflow.com/your-model/42?\
api_key=YOUR_KEY&\
image=https%3A%2F%2Fi.imgur.com%2FPEEvqPN.png"
Windows
설치가 필요합니다 Windows용 curl 와 Windows용 GNU의 base64 도구입니다. 가장 쉬운 방법은 Windows용 git 설치 프로그램을 사용하는 것입니다 여기에는 curl
와 base64
설치 중 "Use Git and optional Unix tools from the Command Prompt"를 선택하면 명령줄 도구도 포함됩니다.
그런 다음 위와 동일한 명령어를 사용할 수 있습니다.
응답 객체 형식
호스팅된 API 추론 경로는 JSON
예측 배열이 포함된 객체를 반환합니다. 각 예측에는 다음 속성이 있습니다:
segmentation_mask
= 입력 이미지와 동일한 크기의 단일 채널 이미지를 base64로 인코딩한 값이며, 각 픽셀 값은 클래스 ID를 나타냅니다.class_map
= 클래스 ID를 클래스 이름에 매핑하는 객체이미지
= 입력 이미지의 크기를 가진 객체height = 입력 이미지의 높이(픽셀 수)
width = 입력 이미지의 너비(픽셀 수)
// 예시 JSON 객체
{
"segmentation_mask": "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",
"class_map": {
"0": " background",
"1": " object"
},
"image": {
"width": 1232,
"height": 821
}
}
API Reference
Inference API 사용하기
POST
https://segment.roboflow.com/:datasetSlug/:versionNumber
base64로 인코딩된 이미지를 직접 모델 엔드포인트에 POST할 수 있습니다. 또는 이미 다른 곳에 호스팅된 이미지라면, 이미지
쿼리 문자열의 파라미터로 URL을 전달할 수 있습니다.
경로 파라미터
datasetSlug
string
데이터셋 이름의 url-safe 버전입니다. 웹 UI의 메인 프로젝트 뷰에서 URL을 확인하거나, 모델 학습 후 데이터셋 버전의 학습 결과 섹션에서 "Get curl command" 버튼을 클릭하여 찾을 수 있습니다.
version
number
데이터셋 버전을 식별하는 버전 번호입니다.
쿼리 파라미터
이미지
string
추가할 이미지의 URL입니다. 이미지가 외부에 호스팅된 경우 사용하세요. (요청 본문에 base64 인코딩 이미지를 POST하지 않을 때 필수) 참고: URL 인코딩을 잊지 마세요.
confidence
number
반환된 예측의 임계값(0-100 기준)입니다. 낮은 숫자는 더 많은 예측을 반환하고, 높은 숫자는 더 적지만 높은 확신의 예측만 반환합니다. 기본값: 50
api_key
string
API 키 (워크스페이스 API 설정 페이지에서 획득)
요청 본문
string
base64로 인코딩된 이미지입니다. (쿼리 파라미터에 이미지 URL을 전달하지 않을 때 필수)
{
"segmentation_mask": "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",
"class_map": {
"0": " background",
"1": " object"
},
"image": {
"width": 1232,
"height": 821
}
}
Last updated
Was this helpful?