pythonlessons

===================================================================================
                                Kernel Shape      Output Shape    Params  \
Layer
0_rb1.convb1.Conv2d_conv       [3, 16, 3, 3]  [1, 16, 32, 128]     448.0
1_rb1.convb1.BatchNorm2d_bn             [16]  [1, 16, 32, 128]      32.0
2_rb1.LeakyReLU_act1                       -  [1, 16, 32, 128]         -
3_rb1.convb2.Conv2d_conv      [16, 16, 3, 3]  [1, 16, 32, 128]     2.32k
4_rb1.convb2.BatchNorm2d_bn             [16]  [1, 16, 32, 128]      32.0
5_rb1.Conv2d_shortcut          [3, 16, 1, 1]  [1, 16, 32, 128]      64.0
6_rb1.LeakyReLU_act2                       -  [1, 16, 32, 128]         -

pythonlessons

import cv2
import typing
import numpy as np

from mltu.inferenceModel import OnnxInferenceModel
from mltu.utils.text_utils import ctc_decoder, get_cer

class ImageToWordModel(OnnxInferenceModel):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)

pythonlessons

Serving TensorBoard on localhost; to expose to the network, use a proxy or pass --bind_all
TensorBoard 2.10.1 at http://localhost:6006/ (Press CTRL+C to quit)

pythonlessons

Tensorboard --logdir Models\08_handwriting_recognition_torch\202303142139\logs

pythonlessons

# create callbacks
earlyStopping = EarlyStopping(monitor='val_CER', patience=20, mode="min", verbose=1)
modelCheckpoint = ModelCheckpoint(configs.model_path + '/model.pt', monitor='val_CER', mode="min", save_best_only=True, verbose=1)
tb_callback = TensorBoard(configs.model_path + '/logs')
reduce_lr = ReduceLROnPlateau(monitor='val_CER', factor=0.9, patience=10, verbose=1, mode='min', min_lr=1e-6)
model2onnx = Model2onnx(
    saved_model_path=configs.model_path + '/model.pt', 
    input_shape=(1, configs.height, configs.width, 3), 
    verbose=1,
    metadata={"vocab": configs.vocab}

pythonlessons

# put on cuda device if available
if torch.cuda.is_available():
    network = network.cuda()

pythonlessons

# uncomment to print network summary, torchsummaryX package is required
summary(network, torch.zeros((1, configs.height, configs.width, 3)))

pythonlessons

loss = CTCLoss(blank=len(configs.vocab))
optimizer = optim.Adam(network.parameters(), lr=configs.learning_rate)

pythonlessons

network = Network(len(configs.vocab), activation='leaky_relu', dropout=0.3)

pythonlessons

# Augment training data with random brightness, rotation and erode/dilate
train_dataProvider.augmentors = [
    RandomBrightness(), 
    RandomErodeDilate(),
    RandomSharpen(),
    RandomRotate(angle=10), 
    ]