TABLE 1.

The convolutional neural network for facial expression recognition.

Convolution block	Output shape	Layer
conv2d	48 × 48 × 64	batch_normalization
		Activation
conv2d_1	24 × 24 × 64	batch_normalization_1
		activation_1
		max_pooling2d
conv2d_2	24 × 24 × 128	batch_normalization_2
		activation_2
conv2d_3	12 × 12 × 128	batch_normalization_3
		activation_3
		max_pooling2d_1
conv2d_4	12 × 12 × 256	batch_normalization_4
		activation_4
conv2d_5	6 × 6 × 256	batch_normalization_5
		activation_5
		max_pooling2d_2
conv2d_6	6 × 6 × 512	batch_normalization_6
		activation_6
conv2d_7	3 × 3 × 512	batch_normalization_7
		activation_7
		max_pooling2d_3
	1 × 1 × 7	dropout, 2-d fc, softmax