Model need Tweaking

OCR/BackendOCR.py

@@ -3,6 +3,11 @@ os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
 import tensorflow as tf
 from tensorflow import keras
 from tensorflow.keras import layers
+from tensorflow.keras.models import Sequential
+from tensorflow.keras.layers import Dense
+from matplotlib import pyplot as plt
+
+
 
 #It hide INFO/Warning and Error, as the usually one will be tensorflow reminding me of not having a cuba driver or the fact it has been installed correctly
 #both doesnt stop the program completing
@@ -13,13 +18,71 @@ batch_size = 10
 train_ds, val_ds = tf.keras.utils.image_dataset_from_directory(
     "ImageTest",
     labels="inferred",
+    label_mode="categorical",
     image_size=image_size,
     batch_size=batch_size,
     validation_split=0.1,
     subset="both",
-    seed=1219
+    seed=1219,
+    color_mode="grayscale"
+)
+
+data_augmentation = keras.Sequential(
+    [
+        layers.RandomFlip("horizontal"),
+        layers.RandomRotation(0.1),
+    ]
+)
+
+train_ds = train_ds.map(
+    lambda img, label: (data_augmentation(img), label),
+    num_parallel_calls=tf.data.AUTOTUNE,
+)
+
+# Prefetching samples in GPU memory helps maximize GPU utilization.
+train_ds = train_ds.prefetch(tf.data.AUTOTUNE)
+val_ds = val_ds.prefetch(tf.data.AUTOTUNE)
+
+model = Sequential()
+
+model.add(Dense(400, activation='relu'))
+model.add(layers.Rescaling(1/255))
+model.add(Dense(200, activation='relu'))
+
+model.add(Dense(500, activation='relu'))
+
+model.add(Dense(200, activation='relu'))
+model.add(layers.Flatten())
+model.add(Dense(10, activation='softmax'))
+
+
+epochs = 40
+
+
+model.compile(
+    optimizer=keras.optimizers.Adam(1e-3),
+    loss="binary_crossentropy",
+    metrics=["accuracy"],
+)
+model.fit(
+    train_ds,
+    epochs=epochs,
+    validation_data=val_ds,
 )
 
+img = keras.utils.load_img(
+    "Test2.jpg", 
+    target_size=image_size,
+    color_mode="grayscale"
+)
+img_array = keras.utils.img_to_array(img)
+img_array = tf.expand_dims(img_array, 0)  # Create batch axis
+
+predictions = model.predict(img_array)
+for v in range(10):
+    print(str(v) + " possible à {0:.2f}".format(predictions[0][v] * 100))
+
 #https://keras.io/examples/vision/image_classification_from_scratch/
 #https://keras.io/examples/
 #https://www.sitepoint.com/keras-digit-recognition-tutorial/
+#https://www.tensorflow.org/api_docs/python/tf/keras/utils/image_dataset_from_directory
\ No newline at end of file