Table 8.

The results of different pre-trained models as well as Tavakoli et al.⁵⁰ on the test-A dataset.

Methods	Lymph			Mono			Neut			Eosi			Baso			Acc (%)
	P (%)	S (%)	F1 (%)	P (%)	S (%)	F1 (%)	P (%)	S (%)	F1 (%)	P (%)	S (%)	F1 (%)	P (%)	S (%)	F1 (%)
ResNet1844	98.84	99.23	99.08	96.96	95.30	96.12	99.66	99.36	99.5	95.77	98.45	97.09	100	100	100	99.06
ResNet3444	98.66	99.52	99.09	96.93	94.44	95.67	99.85	99.14	99.49	94.67	99.38	96.97	100	100	100	99.01
ResNet5044	98.47	99.52	98.99	97.36	94.44	95.88	99.74	99.40	99.57	96.94	98.45	97.69	100	100	100	99.10
ResNext5045	98.01	100	98.99	99.53	91.45	95.32	99.74	99.55	99.64	97.85	98.76	98.30	100	100	100	99.17
MnasNet148	98.93	98.65	98.79	94.14	96.15	95.14	99.66	98.76	99.21	92.15	98.45	95.20	100	100	100	98.59
MobileNet-V247	98.85	99.32	99.08	96.93	94.44	95.67	99.66	99.40	99.53	96.97	99.38	98.16	100	100	100	99.12
DenseNet12146	98.38	99.61	98.99	97.72	91.45	94.48	99.70	99.14	99.42	94.40	99.38	96.82	100	100	100	98.87
ShuffleNet-V249	98.09	99.32	98.70	96.49	94.02	95.24	99.74	99.36	99.55	97.85	98.76	98.30	100	100	100	99.03
VGG1643	98.26	98.26	98.26	95.09	91.03	93.01	99.43	98.57	99	89.01	98.14	93.35	100	100	100	98.09
Tavakoli et al.50	97.23	95.07	96.14	84.87	86.32	85.59	98	95.60	96.78	72.24	91.30	80.66	96.59	95.51	96.05	94.65