francois/HowBadIsMyBatch
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactoring

Browse Source
This commit is contained in:
frankknoll
2023-03-15 17:14:24 +01:00
parent 6330b7b724
commit a9e7bf4833
7 changed files with 68 additions and 82 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/VAERSFileDownloader.py
View File

@@ -4,6 +4,7 @@ from WebDriver import getWebDriver, isCaptchaSolved, saveCaptchaImageAs
from selenium.webdriver.common.by import By from selenium.webdriver.common.by import By
from captcha.CaptchaReader import CaptchaReader from captcha.CaptchaReader import CaptchaReader
from zipUtils import unzipAndRemove from zipUtils import unzipAndRemove
from captcha.CaptchaShape import CaptchaShape
#def getTextInCaptchaImage(captchaImageFile): #def getTextInCaptchaImage(captchaImageFile):
@@ -22,7 +23,8 @@ def solveCaptchaAndStartFileDownload(driver, captchaImageFile):
def _createCaptchaReader(): def _createCaptchaReader():
working_directory = os.path.dirname(__file__) working_directory = os.path.dirname(__file__)
return CaptchaReader(modelFilepath = f'{working_directory}/captcha/MobileNetV3Small') return CaptchaReader(modelFilepath = f'{working_directory}/captcha/MobileNetV3Small',
captchaShape = CaptchaShape())
def downloadFile(absoluteFile, driver, maxTries): def downloadFile(absoluteFile, driver, maxTries):
def _downloadFile(): def _downloadFile():

108
src/captcha/captcha.ipynb → src/captcha.ipynb
View File

@@ -25,22 +25,11 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 1, "execution_count": null,
"metadata": { "metadata": {
"id": "zZSwQragIS_v" "id": "zZSwQragIS_v"
}, },
"outputs": [ "outputs": [],
{
"name": "stderr",
"output_type": "stream",
"text": [
"2023-03-15 10:46:02.303787: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: SSE4.1 SSE4.2 AVX AVX2 FMA\n",
"To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
"/home/frankknoll/.local/lib/python3.9/site-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.16.5 and <1.23.0 is required for this version of SciPy (detected version 1.23.5\n",
" warnings.warn(f\"A NumPy version >={np_minversion} and <{np_maxversion}\"\n"
]
}
],
"source": [ "source": [
"import os\n", "import os\n",
"import numpy as np\n", "import numpy as np\n",
@@ -54,29 +43,29 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 2, "execution_count": null,
"metadata": { "metadata": {
"id": "QB8QZJPg3MGI" "id": "QB8QZJPg3MGI"
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"from GoogleDriveManager import GoogleDriveManager" "from captcha.GoogleDriveManager import GoogleDriveManager"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 3, "execution_count": null,
"metadata": { "metadata": {
"id": "C3bxU1US2blM" "id": "C3bxU1US2blM"
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"from CaptchaGenerator import CaptchaGenerator" "from captcha.CaptchaGenerator import CaptchaGenerator"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 4, "execution_count": null,
"metadata": { "metadata": {
"id": "0DZfMrbe3MGN" "id": "0DZfMrbe3MGN"
}, },
@@ -91,18 +80,18 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 5, "execution_count": null,
"metadata": { "metadata": {
"id": "sNJjugG83MGO" "id": "sNJjugG83MGO"
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"from CharNumConverter import CharNumConverter" "from captcha.CharNumConverter import CharNumConverter"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 6, "execution_count": null,
"metadata": { "metadata": {
"id": "qxs04OTR3MGP" "id": "qxs04OTR3MGP"
}, },
@@ -137,18 +126,18 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 7, "execution_count": null,
"metadata": { "metadata": {
"id": "dAAACymS3MGR" "id": "dAAACymS3MGR"
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"from DatasetFactory import DatasetFactory" "from captcha.DatasetFactory import DatasetFactory"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 8, "execution_count": null,
"metadata": { "metadata": {
"id": "kdL9_t03Mf3t" "id": "kdL9_t03Mf3t"
}, },
@@ -170,7 +159,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 9, "execution_count": null,
"metadata": { "metadata": {
"id": "FqVSEuZp3MGT" "id": "FqVSEuZp3MGT"
}, },
@@ -197,7 +186,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 10, "execution_count": null,
"metadata": { "metadata": {
"id": "apkeCHhP3MGU" "id": "apkeCHhP3MGU"
}, },
@@ -222,18 +211,18 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 11, "execution_count": null,
"metadata": { "metadata": {
"id": "st13jAjL3MGV" "id": "st13jAjL3MGV"
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"from ModelFactory import ModelFactory" "from captcha.ModelFactory import ModelFactory"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 12, "execution_count": null,
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
@@ -244,29 +233,29 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 13, "execution_count": null,
"metadata": { "metadata": {
"id": "B7GZlk2_3MGX" "id": "B7GZlk2_3MGX"
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"from PredictionsDecoder import PredictionsDecoder" "from captcha.PredictionsDecoder import PredictionsDecoder"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 14, "execution_count": null,
"metadata": { "metadata": {
"id": "8Oa7avYt3MGX" "id": "8Oa7avYt3MGX"
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"from ModelDAO import ModelDAO" "from captcha.ModelDAO import ModelDAO"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 15, "execution_count": null,
"metadata": { "metadata": {
"id": "S3X_SslH3MGY" "id": "S3X_SslH3MGY"
}, },
@@ -293,7 +282,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 16, "execution_count": null,
"metadata": { "metadata": {
"id": "NZrKXF6P3MGY" "id": "NZrKXF6P3MGY"
}, },
@@ -304,7 +293,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 17, "execution_count": null,
"metadata": { "metadata": {
"id": "7EsmTaF03MGZ" "id": "7EsmTaF03MGZ"
}, },
@@ -316,7 +305,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 18, "execution_count": null,
"metadata": { "metadata": {
"id": "S_4hl4S4BmZK" "id": "S_4hl4S4BmZK"
}, },
@@ -329,7 +318,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 19, "execution_count": null,
"metadata": { "metadata": {
"id": "WmUghcQaMf3y" "id": "WmUghcQaMf3y"
}, },
@@ -340,28 +329,18 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 20, "execution_count": null,
"metadata": { "metadata": {
"id": "cpxO7yGAMf3z" "id": "cpxO7yGAMf3z"
}, },
"outputs": [ "outputs": [],
{
"name": "stderr",
"output_type": "stream",
"text": [
"2023-03-15 10:41:54.085280: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: SSE4.1 SSE4.2 AVX AVX2 FMA\n",
"To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
"2023-03-15 10:41:54.089954: I tensorflow/core/common_runtime/process_util.cc:146] Creating new thread pool with default inter op setting: 2. Tune using inter_op_parallelism_threads for best performance.\n"
]
}
],
"source": [ "source": [
"charNumConverter = CharNumConverter(CaptchaGenerator.characters)" "charNumConverter = CharNumConverter(CaptchaGenerator.characters)"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 21, "execution_count": null,
"metadata": { "metadata": {
"id": "tVb5nDFTMf3z" "id": "tVb5nDFTMf3z"
}, },
@@ -372,24 +351,23 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 22, "execution_count": null,
"metadata": { "metadata": {},
"id": "t1wzlHQ-Mf3z"
},
"outputs": [], "outputs": [],
"source": [ "source": [
"(img_width, img_height) = (241, 62)" "from captcha.CaptchaShape import CaptchaShape\n",
"captchaShape = CaptchaShape()"
] ]
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 23, "execution_count": null,
"metadata": { "metadata": {
"id": "s35OUslsMf30" "id": "s35OUslsMf30"
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"datasetFactory = DatasetFactory(img_height, img_width, charNumConverter.char_to_num, batch_size = 64)" "datasetFactory = DatasetFactory(captchaShape, charNumConverter.char_to_num, batch_size = 64)"
] ]
}, },
{ {
@@ -403,7 +381,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 24, "execution_count": null,
"metadata": { "metadata": {
"id": "oRcemcbG3MGa" "id": "oRcemcbG3MGa"
}, },
@@ -417,7 +395,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 25, "execution_count": null,
"metadata": { "metadata": {
"id": "P7myCt7e2h6A" "id": "P7myCt7e2h6A"
}, },
@@ -431,7 +409,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 27, "execution_count": null,
"metadata": { "metadata": {
"id": "j9apYsyI3MGb" "id": "j9apYsyI3MGb"
}, },
@@ -474,7 +452,7 @@
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"modelFactory = ModelFactory(img_height, img_width, charNumConverter.char_to_num)" "modelFactory = ModelFactory(captchaShape, charNumConverter.char_to_num)"
] ]
}, },
{ {
@@ -590,7 +568,9 @@
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"model = modelDAO.loadModel(modelName)\n", "# FK-TODO: DRY with VAERSFileDownloader\n",
"modelFilepath = f'{os.getcwd()}/captcha/{modelName}'\n",
"model = modelDAO.loadModel(modelFilepath)\n",
"model.summary(show_trainable=True)" "model.summary(show_trainable=True)"
] ]
}, },
@@ -637,7 +617,7 @@
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"train_dataset, validation_dataset, test_dataset = getTrainValidationTestDatasets(Path(\"captchas/VAERS/\"), datasetFactory)" "train_dataset, validation_dataset, test_dataset = getTrainValidationTestDatasets(Path(\"captcha/captchas/VAERS/\"), datasetFactory)"
] ]
}, },
{ {

10
src/captcha/CaptchaReader.py
View File

@@ -7,21 +7,19 @@ from captcha.DatasetFactory import DatasetFactory
import numpy as np import numpy as np
from tensorflow import keras from tensorflow import keras
# FK-TODO: DRY with captcha.ipynb
img_width = 241
img_height = 62
class CaptchaReader: class CaptchaReader:
def __init__(self, modelFilepath): def __init__(self, modelFilepath, captchaShape):
self.modelFilepath = modelFilepath self.modelFilepath = modelFilepath
self.captchaShape = captchaShape
def getTextInCaptchaImage(self, captchaImageFile): def getTextInCaptchaImage(self, captchaImageFile):
# FK-TODO: refactor
modelDAO = ModelDAO(inColab = False) modelDAO = ModelDAO(inColab = False)
model = modelDAO.loadModel(self.modelFilepath) model = modelDAO.loadModel(self.modelFilepath)
prediction_model = ModelFactory.createPredictionModel(model) prediction_model = ModelFactory.createPredictionModel(model)
charNumConverter = CharNumConverter(CaptchaGenerator.characters) charNumConverter = CharNumConverter(CaptchaGenerator.characters)
datasetFactory = DatasetFactory(img_height, img_width, charNumConverter.char_to_num, batch_size = 64) datasetFactory = DatasetFactory(self.captchaShape,charNumConverter.char_to_num, batch_size = 64)
batchImages = self._asSingleSampleBatch(datasetFactory._encode_single_sample(captchaImageFile, 'dummy')['image']) batchImages = self._asSingleSampleBatch(datasetFactory._encode_single_sample(captchaImageFile, 'dummy')['image'])
preds = prediction_model.predict(batchImages) preds = prediction_model.predict(batchImages)
predictionsDecoder = PredictionsDecoder(CaptchaGenerator.captchaLength, charNumConverter.num_to_char) predictionsDecoder = PredictionsDecoder(CaptchaGenerator.captchaLength, charNumConverter.num_to_char)

5
src/captcha/CaptchaReaderTest.py
View File

@@ -1,5 +1,6 @@
import unittest import unittest
from captcha.CaptchaReader import CaptchaReader from captcha.CaptchaReader import CaptchaReader
from captcha.CaptchaShape import CaptchaShape
import os import os
class CaptchaReaderTest(unittest.TestCase): class CaptchaReaderTest(unittest.TestCase):
@@ -10,7 +11,9 @@ class CaptchaReaderTest(unittest.TestCase):
def test_getTextInCaptchaImage(self): def test_getTextInCaptchaImage(self):
# Given # Given
textInCaptchaImage = '1Ad47a' textInCaptchaImage = '1Ad47a'
captchaReader = CaptchaReader(modelFilepath = f'{self.working_directory}/MobileNetV3Small') captchaReader = CaptchaReader(
modelFilepath = f'{self.working_directory}/MobileNetV3Small',
captchaShape = CaptchaShape())
# When # When
textInCaptchaImageActual = captchaReader.getTextInCaptchaImage(f'{self.working_directory}/captchas/VAERS/{textInCaptchaImage}.jpeg') textInCaptchaImageActual = captchaReader.getTextInCaptchaImage(f'{self.working_directory}/captchas/VAERS/{textInCaptchaImage}.jpeg')

5
src/captcha/CaptchaShape.py Normal file
View File

@@ -0,0 +1,5 @@
class CaptchaShape:
def __init__(self):
self.width = 241
self.height = 62

7
src/captcha/DatasetFactory.py
View File

@@ -3,9 +3,8 @@ import tensorflow as tf
class DatasetFactory: class DatasetFactory:
def __init__(self, img_height, img_width, char_to_num, batch_size): def __init__(self, captchaShape, char_to_num, batch_size):
self.img_height = img_height self.captchaShape = captchaShape
self.img_width = img_width
self.char_to_num = char_to_num self.char_to_num = char_to_num
self.batch_size = batch_size self.batch_size = batch_size
@@ -18,7 +17,7 @@ class DatasetFactory:
def _encode_single_sample(self, img_path, label): def _encode_single_sample(self, img_path, label):
img = tf.io.read_file(img_path) img = tf.io.read_file(img_path)
img = tf.io.decode_jpeg(img, channels=3) img = tf.io.decode_jpeg(img, channels=3)
img = tf.image.resize(img, [self.img_height, self.img_width]) img = tf.image.resize(img, [self.captchaShape.height, self.captchaShape.width])
# Map the characters in label to numbers # Map the characters in label to numbers
label = self.char_to_num(tf.strings.unicode_split(label, input_encoding="UTF-8")) label = self.char_to_num(tf.strings.unicode_split(label, input_encoding="UTF-8"))
# Return a dict as our model is expecting two inputs # Return a dict as our model is expecting two inputs

7
src/captcha/ModelFactory.py
View File

@@ -9,9 +9,8 @@ class ModelFactory:
predictionModelInputLayerName = "image" predictionModelInputLayerName = "image"
predictionModelOutputLayerName = "dense2" predictionModelOutputLayerName = "dense2"
def __init__(self, img_height, img_width, char_to_num): def __init__(self, captchaShape, char_to_num):
self.img_height = img_height self.captchaShape = captchaShape
self.img_width = img_width
self.char_to_num = char_to_num self.char_to_num = char_to_num
# see https://www.tensorflow.org/api_docs/python/tf/keras/applications/resnet/ResNet101 # see https://www.tensorflow.org/api_docs/python/tf/keras/applications/resnet/ResNet101
@@ -52,7 +51,7 @@ class ModelFactory:
def _createModel(self, baseModelFactory, preprocess_input, name): def _createModel(self, baseModelFactory, preprocess_input, name):
# Inputs to the model # Inputs to the model
input_image = layers.Input( input_image = layers.Input(
shape=(self.img_height, self.img_width, 3), shape = (self.captchaShape.height, self.captchaShape.width, 3),
name = ModelFactory.predictionModelInputLayerName, name = ModelFactory.predictionModelInputLayerName,
dtype = "float32") dtype = "float32")
labels = layers.Input(name="label", shape=(None,), dtype="float32") labels = layers.Input(name="label", shape=(None,), dtype="float32")