EasyOCR is a python module for extracting text from image. It is a general OCR that can read both natural scene text and dense text in document. We are currently supporting 80+ languages and expanding.
Fine tunning currently model
-
Download Model for fine tunning : english_g2 -> unzip and move to destination directory all_data
-
Download sample dataset : en_sample -> unzip and move to destination directory saved_models
-
Add config yaml in directory config_files/name_config.yaml I used the filename en_fine_tunning_config.yaml
number: '0123456789' symbol: "!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~ €" lang_char: 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz' experiment_name: 'en_sample' train_data: 'all_data' valid_data: 'all_data/en_sample' manualSeed: 1111 workers: 6 batch_size: 32 #32 num_iter: 300000 # you can custom num_iter valInterval: 100 # you can custom interval validation saved_model: 'saved_models/english_g2.pth' #'saved_models/en_filtered/iter_300000.pth' FT: False optim: False # default is Adadelta lr: 1. beta1: 0.9 rho: 0.95 eps: 0.00000001 grad_clip: 5 #Data processing select_data: 'en_sample' # this is dataset folder in train_data batch_ratio: '1' total_data_usage_ratio: 1.0 batch_max_length: 34 imgH: 64 imgW: 600 rgb: False contrast_adjust: False sensitive: True PAD: True contrast_adjust: 0.0 data_filtering_off: False # Model Architecture Transformation: 'None' FeatureExtraction: 'VGG' SequenceModeling: 'BiLSTM' Prediction: 'CTC' num_fiducial: 20 input_channel: 1 output_channel: 256 hidden_size: 256 decode: 'greedy' new_prediction: False freeze_FeatureFxtraction: False freeze_SequenceModeling: False -
Start training
def get_config(file_path): with open(file_path, 'r', encoding="utf8") as stream: opt = yaml.safe_load(stream) opt = AttrDict(opt) if opt.lang_char == 'None': characters = '' for data in opt['select_data'].split('-'): csv_path = os.path.join(opt['train_data'], data, 'labels.csv') df = pd.read_csv(csv_path, sep='^([^,]+),', engine='python', usecols=['filename', 'words'], keep_default_na=False) all_char = ''.join(df['words']) characters += ''.join(set(all_char)) characters = sorted(set(characters)) opt.character= ''.join(characters) else: opt.character = opt.number + opt.symbol + opt.lang_char os.makedirs(f'./saved_models/{opt.experiment_name}', exist_ok=True) return opt opt = get_config("config_files/en_fine_tunning_config.yaml") train(opt, amp=False)Log Training
training time: 26.636059284210205
[100/300000] Train loss: 0.01502, Valid loss: 0.00408, Elapsed_time: 26.64084
Current_accuracy : 99.320, Current_norm_ED : 0.9993
Best_accuracy : 99.320, Best_norm_ED : 0.9993
--------------------------------------------------------------------------------
Ground Truth | Prediction | Confidence Score & T/F
--------------------------------------------------------------------------------
Zsasz @Johnnie Dean_DMSP | Zsasz @Johnnie Dean_DMSP | 0.0232 True
Rupp ? OCCUPIED Wmkx | Rupp ? OCCUPIED Wmkx | 0.3271 True
--------------------------------------------------------------------------------
-
Run model training in easy_ocr
- model: easyocr default saved in ~.EasyOCR, please move your model trained in ~.EasyOCR/model/en_sample.pth
- user_network: in directory user_network there are neural network file and config. Please create a files
- user network : filename en_sample.py -> you can custom filename
import torch.nn as nn class BidirectionalLSTM(nn.Module): def __init__(self, input_size, hidden_size, output_size): super(BidirectionalLSTM, self).__init__() self.rnn = nn.LSTM(input_size, hidden_size, bidirectional=True, batch_first=True) self.linear = nn.Linear(hidden_size * 2, output_size) def forward(self, input): """ input : visual feature [batch_size x T x input_size] output : contextual feature [batch_size x T x output_size] """ try: # multi gpu needs this self.rnn.flatten_parameters() except: # quantization doesn't work with this pass recurrent, _ = self.rnn(input) # batch_size x T x input_size -> batch_size x T x (2*hidden_size) output = self.linear(recurrent) # batch_size x T x output_size return output class VGG_FeatureExtractor(nn.Module): def __init__(self, input_channel, output_channel=256): super(VGG_FeatureExtractor, self).__init__() self.output_channel = [int(output_channel / 8), int(output_channel / 4), int(output_channel / 2), output_channel] self.ConvNet = nn.Sequential( nn.Conv2d(input_channel, self.output_channel[0], 3, 1, 1), nn.ReLU(True), nn.MaxPool2d(2, 2), nn.Conv2d(self.output_channel[0], self.output_channel[1], 3, 1, 1), nn.ReLU(True), nn.MaxPool2d(2, 2), nn.Conv2d(self.output_channel[1], self.output_channel[2], 3, 1, 1), nn.ReLU(True), nn.Conv2d(self.output_channel[2], self.output_channel[2], 3, 1, 1), nn.ReLU(True), nn.MaxPool2d((2, 1), (2, 1)), nn.Conv2d(self.output_channel[2], self.output_channel[3], 3, 1, 1, bias=False), nn.BatchNorm2d(self.output_channel[3]), nn.ReLU(True), nn.Conv2d(self.output_channel[3], self.output_channel[3], 3, 1, 1, bias=False), nn.BatchNorm2d(self.output_channel[3]), nn.ReLU(True), nn.MaxPool2d((2, 1), (2, 1)), nn.Conv2d(self.output_channel[3], self.output_channel[3], 2, 1, 0), nn.ReLU(True)) def forward(self, input): return self.ConvNet(input) class Model(nn.Module): def __init__(self, input_channel, output_channel, hidden_size, num_class): super(Model, self).__init__() """ FeatureExtraction """ self.FeatureExtraction = VGG_FeatureExtractor(input_channel, output_channel) self.FeatureExtraction_output = output_channel self.AdaptiveAvgPool = nn.AdaptiveAvgPool2d((None, 1)) """ Sequence modeling""" self.SequenceModeling = nn.Sequential( BidirectionalLSTM(self.FeatureExtraction_output, hidden_size, hidden_size), BidirectionalLSTM(hidden_size, hidden_size, hidden_size)) self.SequenceModeling_output = hidden_size """ Prediction """ self.Prediction = nn.Linear(self.SequenceModeling_output, num_class) def forward(self, input, text): """ Feature extraction stage """ visual_feature = self.FeatureExtraction(input) visual_feature = self.AdaptiveAvgPool(visual_feature.permute(0, 3, 1, 2)) visual_feature = visual_feature.squeeze(3) """ Sequence modeling stage """ contextual_feature = self.SequenceModeling(visual_feature) """ Prediction stage """ prediction = self.Prediction(contextual_feature.contiguous()) return prediction-
config : filename en_sample.yaml -> you can custom filename
network_params: input_channel: 1 output_channel: 256 hidden_size: 256 imgH: 64 lang_list: - 'en' character_list: 0123456789!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~ €ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz -
Interface load model
import easyocr reader = easyocr.Reader(['en'], recog_network='en_sample')