A New Image Classification Approach via Improved MobileNet Models with Local Receptive Field Expansion in Shallow Layers

Table 2

Classification accuracy rates (%) on Caltech-101 dataset.

Number of iterations	30000	35000	40000	45000	50000
SqueezeNet	53.60	53.60	53.47	53.40	53.47
MobileNets	76.73	76.60	76.60	76.80	76.60
Dense1-MobileNet	76.60	76.53	76.47	76.40	76.47
Dense2-MobileNet	77.60	77.67	77.87	77.80	77.80
Dilated1-MobileNet	77.40	77.47	77.53	77.40	77.47
Dilated2-MobileNet	77.67	77.80	77.73	77.67	77.73
Dilated3-MobileNet	78.60	78.60	78.53	78.53	78.73