Online English Teaching Quality Assessment Based on K-Means and Improved SSD Algorithm

Table 1

Network layer parameters of improved DenseNet.

Network layer name	Output size per pixel	Image manipulation
Convolution 1	64 × 150 × 150	3 × 3 conv, stride 2
Convolution 2	64 × 150 × 150	3 × 3 conv, stride 1
Convolution 3	128 × 150 × 150	3 × 3 conv, stride 1
Pooling	128 × 75 × 75	2 × 2 max pool, stride 2
Dense block 1	416 × 75 × 75	(1 × 1 conv, 3 × 3 conv) × 6
Dense block 2	800 × 38 × 38	(1 × 1 conv, 3 × 3 conv) × 8
Dense block 3	1,184 × 19 × 19	(1 × 1 conv, 3 × 3 conv) × 8
Dense block 4	1,586 × 19 × 19	(1 × 1 conv, 3 × 3 conv) × 8
Transition layer 1	416 × 75 × 75	1 × 1 conv
	416 × 38 × 38	2 × 2 max pool, stride 2
Transition layer 2	800 × 38 × 38	1 × 1 conv
	800 × 19 × 19	2 × 2 max pool, stride 2
Transition layer 3	1,184 × 19 × 19	1 × 1 conv
Transition layer 4	1,586 × 19 × 19	1 × 1 conv