【PaddlePaddle+OpenVINO】打造一个会发声的电表检测识别器

!pip install paddlehub==2.2.0 -i https://mirror.baidu.com/pypi/simple

Installing dependent packages from /home/aistudio/.paddlehub/tmp/tmpd42h5sba/fastspeech2_baker/requirements.txt: -

!unzip openvino-deploy.zip

%cd openvino-deploy/Parakeet

# 安装parakeet!pip install -e .

[nltk_data] Error loading averaged_perceptron_tagger: [Errno 110]
[nltk_data]     Connection timed out

# nltk_data的国内镜像，很多开发者都同步过，随便找一个就行!git clone https://gitee.com/eurake/nltk_data.git

!mv nltk_data/packages ~/nltk_data

import nltk
nltk.download("punkt")
nltk.download("cmudict")

# 安装FastSpeech2预训练模型!hub install fastspeech2_baker

!mv Parakeet ./openvino-deploy/Parakeet

%cd ~

!tree ./openvino-deploy

# 初始化ttstts = hub.Module(name='fastspeech2_baker', version='1.0.0') 
# 视频流文件video_file = "./data/SVID_20220411_003747_1.mp4"# 启动视频流文件电表读数编号的实时检测run_paddle_ocr(source=video_file, flip=False, use_popup=True, tts=tts)# 调用摄像头进行电表读数编号的实时检测# run_paddle_ocr(source=0, flip=False, use_popup=True, tts=tts)

if rec_res != []:
    image = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
    boxes = dt_boxes    # 只对置信度高于0.5的识别文本进行语音播报
    txts = [rec_res[i][0] for i in range(len(rec_res)) if rec_res[i][1]>0.5] 
    scores = [rec_res[i][1] for i in range(len(rec_res))] 

    # 生成原图和识别结果对比图，分两栏左右并排显示
    draw_img = processing.draw_ocr_box_txt(
        image,
        boxes,
        txts,
        scores,
        drop_score=0.5)    # 可视化OCR识别结果
    _, f_width = draw_img.shape[:2]
    fps = 1000 / processing_time_det
    cv2.putText(img=draw_img, text=f"OpenVINO Inference time: {processing_time_det:.1f}ms ({fps:.1f} FPS)", 
                org=(20, 40),fontFace=cv2.FONT_HERSHEY_COMPLEX, fontScale=f_width / 1000,
                color=(0, 0, 255), thickness=1, lineType=cv2.LINE_AA)    # use this workaround if there is flickering
    if use_popup: 
        draw_img = cv2.cvtColor(draw_img, cv2.COLOR_RGB2BGR)
        cv2.imshow(winname=title, mat=draw_img)
        key = cv2.waitKey(1)        if key == 32:
            cv2.imwrite('test.jpg', draw_img)            for i in range(len(txts)):                if len(txts[i])>8:
                    txts[i] = '电表编号是' + str(txts[i])                else:
                    txts[i] = '电表读数是' + str(txts[i])            print(txts)            if len(txts) > 0:                # 英文模型：'fastspeech2_ljspeech'，中文：fastspeech2_baker
                tts.generate(txts) # 中英文要一致
                # 使用playsound播报识别结果，会有略微卡顿
                for i in range(len(txts)):
                    playsound('./wavs/%s.wav' % str(i+1))        # escape = 27，按ESC键退出
        if key == 27:
            cv2.destroyAllWindows()            break
    else:        # encode numpy array to jpg
        draw_img = cv2.cvtColor(draw_img, cv2.COLOR_RGB2BGR)
        _, encoded_img = cv2.imencode(ext=".jpg", img=draw_img,
                                      params=[cv2.IMWRITE_JPEG_QUALITY, 100])        # create IPython image
        i = display.Image(data=encoded_img)        # display the image in this notebook
        display.clear_output(wait=True)
        display.display(i)

from playsound import playsound
 
playsound('wavs/1.wav')

%cd openvino-deploy

# 该代码须在本地运行!python demo_vino.py

(paddle) C:\MachineLearning\openvino-deploy>python demo_vino.py
C:\Users\noname\.conda\envs\paddle\lib\site-packages\paddle\framework\io.py:415: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated since Python 3.3,and in 3.9 it will stop working  if isinstance(obj, collections.Iterable) and not isinstance(obj, (
[2022-05-03 19:29:20,808] [    INFO] - Load fastspeech2 params from C:\Users\noname\.paddlehub\modules\fastspeech2_baker\assets\fastspeech2_nosil_baker_ckpt_0.4\snapshot_iter_76000.pdz
[2022-05-03 19:29:21,320] [    INFO] - Load vocoder params from C:\Users\noname\.paddlehub\modules\fastspeech2_baker\assets\pwg_baker_ckpt_0.4\pwg_snapshot_iter_400000.pdz
[2022-05-03 19:30:10] [DEBUG] [__init__.py:166] Prefix dict has been built successfully.
[2022-05-03 19:30:16,205] [    INFO] - 1 wave files have been generated in C:\MachineLearning\openvino-deploy\wavs
['电表读数是022809']
[2022-05-03 19:30:24,352] [    INFO] - 1 wave files have been generated in C:\MachineLearning\openvino-deploy\wavs
Source ended
[2022-05-03 19:30:58] [DEBUG] [__init__.py:166] Prefix dict has been built successfully.[2022-05-03 19:31:04,633] [    INFO] - 1 wave files have been generated in C:\MachineLearning\openvino-deploy\wavs
Source ended

import IPython
IPython.display.Audio('./wavs/1.wav')