bridgecrew-perf7 / image-classification-model-deployment-5

Convert Trained Image Classification model into Tflite

Image-Classification-Model-Deployment

Convert Trained Image Classification model into Tflite

Dicoding-Image-Classification-Submission

• Dataset is provided by Kaggle uploaded by Puneet Bansal
• Link to File : https://www.kaggle.com/puneet6060/intel-image-classification
• The dataset can be fetch using kaggle library on colab
• 13.500samples
• 80:20 train test split

Library Used

• Tensorflow
• Keras
• matplotlib
• Numpy
• Pandas
• Kaggle

Data set composition

	type	buildings	forest	mountain	sea	street	total
0	train	2191	2271	2512	2274	2382	11630
1	val	437	474	525	510	501	2447

Step by Step Model Building

Display random images to examine

Rescaling Image

Image Augmenting (train set only)

• Rotation
• Horizontal flip
• Zoom range
• Shear
• Fill mode nearest
• Data generating
• resizing image to 150x150
• batch size 128
• class mode categorical

Build Convnet

• 1 hidden layer (perceptron 128 units)
• output layer 'Softmax'
• model = keras.Sequential([
• layers.Conv2D(32, (3,3), activation = 'relu', input_shape= (150,150,3)),
• layers.MaxPooling2D(pool_size=(2, 2)),
• layers.Conv2D(64,(3,3), activation= 'relu'),
• layers.MaxPooling2D(pool_size=(2, 2)),
• layers.Conv2D(128,(3,3), activation= 'relu'),
• layers.MaxPooling2D(pool_size=(2, 2)),
• layers.Flatten(),
• layers.Dropout(0.5),
• layers.Dense(128, activation= 'relu'),
• layers.Dense(5, activation= 'softmax')
• ])

Adding Loss Function and Optimizer

• loss : categorical cross entropy
• optimizer adam
• metrics accuracy

Adding Custom Early stop function using Callback from tensorflow

• Model will stop training once val accuracy reaches atleast 85%

Training the model

• epoch : 40
• steps per epoch : 20
• verbose : 1
• validation steps : 10
• callbacks

Results

• Best Validation Loss: 0.41

• Best Validation Accuracy: 0.86

• Loss

• Accuracy

Convert Model to Tflite

• Name trainde model to my model.tflite