Handwritten Character Recognition - Training¶

import sys

if 'google.colab' in sys.modules:
    !git clone https://github.com/luidsonl/handwritten-character-recognition.git
    %cd handwritten-character-recognition/training

x_train, y_train, x_test, y_test = load_and_preprocess()

print(f"Training set: {x_train.shape}, {y_train.shape}")
print(f"Test set:     {x_test.shape}, {y_test.shape}")
print(f"Classes:      {y_train.shape[1]}")

_viz_rng = np.random.default_rng(42)

fig, axes = plt.subplots(2, 5, figsize=(12, 5))
for i, ax in enumerate(axes.flat):
    idx = _viz_rng.integers(len(x_train))
    ax.imshow(x_train[idx].squeeze(), cmap='gray')
    ax.set_title(get_label(np.argmax(y_train[idx])))
    ax.axis('off')
plt.tight_layout()
plt.show()

model = build_model()
model = compile_model(model, learning_rate=LEARNING_RATE)
model.summary()

history = train_model(
    model,
    x_train,
    y_train,
    epochs=EPOCHS,
    batch_size=BATCH_SIZE,
    validation_split=VALIDATION_SPLIT,
)

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 4))

ax1.plot(history.history['accuracy'], label='Train')
ax1.plot(history.history['val_accuracy'], label='Validation')
ax1.set_title('Accuracy')
ax1.set_xlabel('Epoch')
ax1.legend()

ax2.plot(history.history['loss'], label='Train')
ax2.plot(history.history['val_loss'], label='Validation')
ax2.set_title('Loss')
ax2.set_xlabel('Epoch')
ax2.legend()

plt.tight_layout()
plt.show()

results = evaluate_model(model, x_test, y_test)
print_evaluation_report(results)

cm = results['confusion_matrix']

plt.figure(figsize=(14, 12))
sns.heatmap(cm, annot=False, fmt='d', cmap='Blues',
            xticklabels=LABELS, yticklabels=LABELS)
plt.xlabel('Predicted')
plt.ylabel('True')
plt.title('Confusion Matrix')
plt.tight_layout()
plt.show()

confused = get_most_confused_pairs(cm, top_n=15)
print("Top 15 confused pairs (true -> predicted):")
print("-" * 40)
for true_label, pred_label, count in confused:
    print(f"  '{true_label}' -> '{pred_label}': {count}")

import os

os.makedirs("models", exist_ok=True)
save_model_keras(model, "models/emnist_model.keras")
print("Model saved to models/emnist_model.keras")

import os
import base64
from IPython.display import display, HTML

if 'google.colab' in sys.modules:
    path = os.path.abspath("models/emnist_model.keras")
    with open(path, "rb") as f:
        data = base64.b64encode(f.read()).decode()
    filename = os.path.basename(path)
    display(HTML(f"""
    <button onclick="document.getElementById('dl-keras').click()"
            style="padding:8px 16px;background:#4CAF50;color:white;border:none;border-radius:4px;cursor:pointer;font-size:14px;">
      Baixar {filename}
    </button>
    <a id="dl-keras" href="data:application/octet-stream;base64,{data}" download="{filename}" style="display:none"></a>
    """))

from libs.model import convert_to_tfjs
import os
import base64
from IPython.display import display, HTML

os.makedirs("web_model", exist_ok=True)
convert_to_tfjs(model, "web_model")

for f in sorted(os.listdir("web_model")):
    size = os.path.getsize(os.path.join("web_model", f))
    print(f"  {f}: {size:,} bytes")

if 'google.colab' in sys.modules:
    buttons_html = ''
    for f in sorted(os.listdir("web_model")):
        path = os.path.abspath(os.path.join("web_model", f))
        with open(path, "rb") as fh:
            data = base64.b64encode(fh.read()).decode()
        btn_id = f"dl-{f.replace('.', '-')}"
        color = '#2196F3' if f.endswith('.json') else '#FF9800'
        buttons_html += f"""
        <button onclick="document.getElementById('{btn_id}').click()"
                style="padding:8px 16px;background:{color};color:white;border:none;border-radius:4px;cursor:pointer;font-size:14px;margin-right:8px;">
          Baixar {f}
        </button>
        <a id="{btn_id}" href="data:application/octet-stream;base64,{data}" download="{f}" style="display:none"></a>
        """
    display(HTML(buttons_html))