PyAutoMakerHuman
사람과 관련된 자동화 모듈
주의
이 프로젝트는 아직 개발중입니다 내보내는 API는 언제든지 바뀔 수 있습니다
설치
pip install PyAutoMakerHuman
또는
git clone https://github.com/boa9448/PyAutoMakerHuman.git
cd PyAutoMakerHuman
python setup.py install
사용
데이터셋 폴더 구조
dataset
├─face
└─hand
├─paper
├─rock
└─scissors
분류기 학습
import os
import PyAutoMakerHuman as pamh
#데이터셋 경로 지정
dataset_path = os.path.join("dataset", "hand")
#손 검출 임계치 0.7
hand = pamh.HandUtil(min_detection_confidence=0.7)
#데이터셋의 경로를 넘겨주면 사전형 값을 리턴함
data = hand.extract_dataset(dataset_path)
train_data, name = data["data"], data["name"]
trainer = pamh.SvmUtil()
#svm 학습
trainer.train_svm(train_data, name)
#학습된 svm 모델을 myModel폴더에 저장
trainer.save_svm("myModel")
이미지에서 손모양 분류
import cv2
import imutils
import os
import PyAutoMakerHuman as pamh
import random
hand = pamh.HandUtil(min_detection_confidence=0.7)
trainer = pamh.SvmUtil()
#학습된 svm모델을 불러옴
#myModel폴더에 있는 모델과 라벨을 불러옴
trainer.load_svm("myModel")
#이미지를 읽어옴
img_path = "test.jpg"
img = cv2.imread(img_path)
#이미지 리사이즈
img = imutils.resize(img, width=600)
#이미지에서 박스와 랜드마크를 추출함
data_list = hand.extract(img)
color_dict = {}
#이미지에서 탐지를 성공했다면...
if data_list:
for idx, data in enumerate(data_list):
landmark_1d = data[1]
name, proba = trainer.predict([landmark_1d])
box = data[0]
color = (0, 0, 0)
if name in color_dict:
color = color_dict[name]
else:
color = (random.randint(0, 255), random.randint(0, 255), random.randint(0, 255))
color_dict[name] = color
cv2.putText(img, f"{name} : {proba:.2f}", (box[0], box[1] - 25)
, cv2.FONT_HERSHEY_SIMPLEX, 1, color, 2, cv2.LINE_AA)
cv2.rectangle(img, (box[0], box[1]), (box[0] + box[2], box[1] + box[3])
, color, 2)
cv2.imshow("view", img)
cv2.waitKey()
cv2.destroyAllWindows()
추후 개발
간단하게 모델을 학습 시킬 수 있는 GUI도구
GUI도구 미리 둘러보기
패키지 설치 후 python -m PyAutoMakerHuman
Metadata
Release files for PyAutoMakerHuman 1.2.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| PyAutoMakerHuman-1.2.1.tar.gz | 47.9 MB | Without provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| PyAutoMakerHuman-1.2.1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 95.7 MB
Release files / PyAutoMakerHuman-1.2.1.tar.gz
| Download URL | PyAutoMakerHuman-1.2.1.tar.gz |
|---|---|
| Size | 47.9 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
55ce2b8c97918298973746074b3ce95fbb2845a6f7ae6116ad51dd5fc3285f69
|
|
BLAKE2b-256 checksum How to use checksums |
3b526047ececbb988012cfdccd5e608410399d630653a38941abd81969aefb86
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/3.4.1 importlib_metadata/4.6.1 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.61.2 CPython/3.9.6
|
Release files / PyAutoMakerHuman-1.2.1-py3-none-any.whl
| Download URL | PyAutoMakerHuman-1.2.1-py3-none-any.whl |
|---|---|
| Size | 47.9 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
fb20641792a4bfef2e9900a189f6c27b3c192c7e5cae5798351bc6c32560283d
|
|
BLAKE2b-256 checksum How to use checksums |
ff3a9b41720dde0638435d04f732ba382e859bc3f25b2fad28bf8e6f3797262a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/3.4.1 importlib_metadata/4.6.1 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.61.2 CPython/3.9.6
|
Security for PyAutoMakerHuman 1.2.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.