私的AI研究会 > OCR_jp
日本語認識(OCR) を検証する。
下記サイトの公開プログラムをテストしてみる。
$ git clone https://github.com/tanreinama/OCR_Japanease.git Cloning into 'OCR_Japanease'... remote: Enumerating objects: 122, done. remote: Counting objects: 100% (122/122), done. remote: Compressing objects: 100% (92/92), done. remote: Total 122 (delta 53), reused 77 (delta 23), pack-reused 0 Receiving objects: 100% (122/122), 3.45 MiB | 19.87 MiB/s, done. Resolving deltas: 100% (53/53), done. $ cd OCR_Japanease $ ls LICENSE nets report testshot1.png-detections.png misc ocr_japanease.py requirements.txt testshot2.png models readme.md testshot1.png testshot2.png-detections.png
$ wget https://nama.ne.jp/models/ocr_jp-v2.zip --2021-04-12 14:00:01-- https://nama.ne.jp/models/ocr_jp-v2.zip nama.ne.jp (nama.ne.jp) をDNSに問いあわせています... 112.78.112.176 nama.ne.jp (nama.ne.jp)|112.78.112.176|:443 に接続しています... 接続しました。 HTTP による接続要求を送信しました、応答を待っています... 200 OK 長さ: 180256769 (172M) [application/zip] `ocr_jp-v2.zip' に保存中 ocr_jp-v2.zip 100%[===================>] 171.91M 10.6MB/s in 17s 2021-04-12 14:00:17 (10.4 MB/s) - `ocr_jp-v2.zip' へ保存完了 [180256769/180256769] $ unzip ocr_jp-v2.zip Archive: ocr_jp-v2.zip inflating: models/detectionnet.model inflating: models/classifiernet.model
mizutu@ubuntu2004dk:~/OCR_Japanease$ python3 ocr_japanease.py testshot1.png
Traceback (most recent call last):
File "ocr_japanease.py", line 11, in <module>
from misc.detection import Detector
File "/home/mizutu/OCR_Japanease/misc/detection.py", line 5, in <module>
from sklearn.cluster import OPTICS
ModuleNotFoundError: No module named 'sklearn'
mizutu@ubuntu2004dk:~/OCR_Japanease$ sudo pip3 install scikit-learn
Collecting scikit-learn
Downloading scikit_learn-0.24.1-cp38-cp38-manylinux2010_x86_64.whl (24.9 MB)
|████████████████████████████████| 24.9 MB 21.4 MB/s
Requirement already satisfied: numpy>=1.13.3 in /usr/local/lib/python3.8/dist-packages (from scikit-learn) (1.18.5)
Collecting joblib>=0.11
Downloading joblib-1.0.1-py3-none-any.whl (303 kB)
|████████████████████████████████| 303 kB 41.3 MB/s
Collecting scipy>=0.19.1
Downloading scipy-1.6.2-cp38-cp38-manylinux1_x86_64.whl (27.2 MB)
|████████████████████████████████| 27.2 MB 27.3 MB/s
Collecting threadpoolctl>=2.0.0
Downloading threadpoolctl-2.1.0-py3-none-any.whl (12 kB)
Installing collected packages: threadpoolctl, scipy, joblib, scikit-learn
Successfully installed joblib-1.0.1 scikit-learn-0.24.1 scipy-1.6.2 threadpoolctl-2.1.0
WARNING: You are using pip version 20.3.3; however, version 21.0.1 is available.
You should consider upgrading via the '/usr/bin/python3 -m pip install --upgrade pip' command.mizutu@ubuntu2004dk:~/OCR_Japanease$ python3 Python 3.8.5 (default, Jan 27 2021, 15:41:15) [GCC 9.3.0] on linux Type "help", "copyright", "credits" or "license" for more information. >>> import sklearn >>> print(sklearn.__version__) 0.24.1 >>> exit()
$ python3 ocr_japanease.py testshot1.png
Traceback (most recent call last):
File "ocr_japanease.py", line 157, in <module>
main()
File "ocr_japanease.py", line 34, in main
ocr_result = get_ocr(d, dpi=args.dpi, use_cuda=(not args.cpu), output_detect_img=args.output_detect_img, low_gpu_memory=args.low_gpu_memory)
File "ocr_japanease.py", line 72, in get_ocr
model.load_state_dict(torch.load(det_model))
File "/home/mizutu/.local/lib/python3.8/site-packages/torch/serialization.py", line 592, in load
return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args)
File "/home/mizutu/.local/lib/python3.8/site-packages/torch/serialization.py", line 851, in _load
result = unpickler.load()
File "/home/mizutu/.local/lib/python3.8/site-packages/torch/serialization.py", line 843, in persistent_load
load_tensor(data_type, size, key, _maybe_decode_ascii(location))
File "/home/mizutu/.local/lib/python3.8/site-packages/torch/serialization.py", line 832, in load_tensor
loaded_storages[key] = restore_location(storage, location)
File "/home/mizutu/.local/lib/python3.8/site-packages/torch/serialization.py", line 175, in default_restore_location
result = fn(storage, location)
File "/home/mizutu/.local/lib/python3.8/site-packages/torch/serialization.py", line 151, in _cuda_deserialize
device = validate_cuda_device(location)
File "/home/mizutu/.local/lib/python3.8/site-packages/torch/serialization.py", line 135, in validate_cuda_device
raise RuntimeError('Attempting to deserialize object on a CUDA '
RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.
$ python3 ocr_japanease.py --cpu testshot1.png file "testshot1.png" detected in 72 dpi. [Block #0] コロナウイルスにまけるな [Block #1] がんばろう [Block #2] 日本