Open Access

Improved Method of ResNet50 Image Classification Based on Transfer Learning

, ,  and   
Jun 16, 2025

Cite
Download Cover

Figure 1.

Residual unit structure diagram
Residual unit structure diagram

Figure 2.

Comparison of accuracy between the two networks during training
Comparison of accuracy between the two networks during training

Figure 3.

Comparison of loss values between the two networks during training
Comparison of loss values between the two networks during training

Figure 4.

Comparison of confusion matrices between the two networks during training
Comparison of confusion matrices between the two networks during training

resnet50 architecture

convolutional layer output layer ResNet50
Conv-1 112×112 7×7, 64, S=2 3×3 maxpool, S=2
Conv2-x 56×56 [ 1×1643×3641×1256 ]*3
Conv3-x 28×28 [ 1×11283×31281×1512 ]*4
Conv4-x 14×14 [ 1×12563×32561×11024 ]*6
Conv5-x 7×7 [ 1×15123×35121×12048 ]*3
1×1 Average_pool,1000-dfc, Soft_max
Flops 3.8×109
Language:
English
Publication timeframe:
4 times per year
Journal Subjects:
Computer Sciences, Computer Sciences, other