Deep Residual Learning for Image Recognition

Kaiming HeXiangyu ZhangShaoqing RenJian Sun

   Papers with code   Abstract  PDF

Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously... (read more)

Benchmarked Models

RANK
MODEL
REPO
CODE RESULT
PAPER RESULT
ε-REPRODUCED
BUILD
1
ResNet-152b
80.1%
--
2
ResNet-101b
79.4%
--
3
ResNet-152
79.0%
78.6%
4
ResNet-152
78.2%
78.6%
5
ResNet-101
78.1%
78.2%
6
ResNet-50
77.7%
77.1%
7
ResNet-50b
77.6%
--
8
ResNet-101
77.3%
78.2%
9
ResNet-BC-38b
76.3%
--
10
ResNet-50
75.9%
77.1%
11
ResNet-34
75.2%
--
12
ResNet-BC-26b
74.9%
--
13
ResNet-26
73.7%
--
14
ResNet-34 A
73.2%
75.0%
15
ResNet-18
71.5%
72.1%
16
ResNet-18 x0.75
69.6%
--
17
ResNet-18
69.5%
72.1%
18
ResNet-16
69.5%
--
19
ResNet-BC-14b
69.3%
--
20
ResNet-14
67.5%
--
21
ResNet-12
66.4%
--
22
ResNet-18 x0.5
66.2%
--
23
ResNet-10
65.3%
--
24
ResNet-18 x0.25
60.4%
--