Metadata-Version: 2.1
Name: indOCRArmy
Version: 0.1.12
Summary: A package for reading id and name on KTP and SIM
Home-page: UNKNOWN
Author: Mulya Rahardja Madjiah, Andreas; Hervind; S Fawwazi, Ziyad
Author-email: ziyad.syauqi95@gmail.com
License: UNKNOWN
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.7
Description-Content-Type: text/markdown
Requires-Dist: requests
Requires-Dist: numpy
Requires-Dist: opencv-contrib-python
Requires-Dist: pandas
Requires-Dist: Pillow
Requires-Dist: matplotlib

# Indo OCR Army

This is a package for OCR Identity card in Indonesia (KTP and SIM). This packages build on top of detectron2 so you should install detectron2 first and some requirements need to install separately before you run this packages, Note that we test it on a device with `GPU/CUDA` if there is an error please report an issue to our email. After install this package, please follow the tutorial bellow:

```
# if use conda
conda install pytorch torchvision cudatoolkit=10.2 -c pytorch

# if use python
pip install torch==1.5.1
pip install torchvision
```

```
python -m pip install detectron2 -f \ https://dl.fbaipublicfiles.com/detectron2/wheels/cu102/torch1.5/index.html
```

for comprehensive use, you can do this:

```
import os
import cv2
import matplotlib.pyplot as plt

from IndoOCRArmy import defaultConfig, DrawOCR, numericalDetectron2, boundingBoxesDetectron2, alphabeticalDetectron2, easypredict

# load classes
cfg = defaultConfig()

for key in cfg.keys():
    if 'list_cuda' in cfg[key]:
        cfg[key]['list_cuda'] = [0]

drawer_ocr = DrawOCR(cfg['drawOCR'])
bBoxDet = boundingBoxesDetectron2(cfg['boundingBoxesDetectron2'])
numDet = numericalDetectron2(cfg['numericalDetectron2'])
alphaDet = alphabeticalDetectron2(cfg['alphabeticalDetectron2'])

# load image
image_ktp = cv2.imread("assets/ktp_example.jpg")
image_sim = cv2.imread("assets/sim_example.jpg")

# detect boundingboxes
crops, boxes, labels = bBoxDet.predict(image_ktp, input_type='ktp')

# detect number and alphabet
dict_ID = numDet.predict_ensemble(crops[0])
dict_Name = alphaDet.predict_ensemble(crops[1])

# choose `weighted_hardvote_word` for the best result according to our benchmark
ID =  dict_ID.get("weighted_hardvote_word")
Name =  dict_Name.get("weighted_hardvote_word")

# parse NIK to get information about : location, gender, and birthdate
parse_NIK = numDet.parse_nik(ID)

# create listdata and listlabel for visualization later
listdata = [ID, Name]
listlabel = [x for x in list(labels.values()) if x is not None]
for label, data in parse_NIK.items():
    listdata.append(data)
    listlabel.append(label)

print(ID)
print(Name)

drawer_ocr.show_list_images(list_img=crops.values())
```

For visuzlize comprehensive result, use this:

```
drawer_ocr.show_desc(image_ktp, boxes, labels, listdata, listlabel)
```

For quick result, use this:
```
image_ktp = cv2.imread("assets/ktp_example.jpg")
easypredict(image_ktp, bBoxDet, numDet, alphaDet, input_type='ktp')
```


