MearcAll papers

arXiv 1512.03385 · cs.CV

Deep Residual Learning for Image Recognition

Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun

Claims were extracted automatically by reducer-1 on 2026-11-12. The paper's license does not allow republishing it, so long claims are shortened to a quotation. Read the full text on arXiv.

No reproduction recordedres_d759d80c
An ensemble of these residual nets achieves 3.57% error on the ImageNettest set.

Abstract · page 1 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_c1f23784
Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset.

Abstract · page 1 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_027ebee5
Unexpectedly, such degradation is not caused by overfitting , and adding more layers to a suitably deep model leads to higher training error, as reported in [11, 42] and thoroughly … (excerpt)

Introduction · page 1 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_fe8c998c
We present comprehensive experiments on ImageNet [36] to show the degradation problem and evaluate our method.

Introduction · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_06c2da64
We show that: 1) Our extremely deep residual nets are easy to optimize, but the counterpart “plain” nets (that simply stack layers) exhibit higher training error when the depth increases; … (excerpt)

Introduction · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_33738518
Our ensemble has 3.57% top-5 error on the ImageNet test set, and won the 1st place in the ILSVRC 2015 classification competition .

Introduction · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_00f66dd2
The results in Table 2 show that the deeper 34-layer plain net has higher validation error than the shallower 18-layer plain net.

Experiments · page 4 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_70d31aee
34-layer plain net has higher training error throughout the whole training procedure, even though the solution space of the 18-layer plain network is a subspace of that of the 34-layer … (excerpt)

Experiments · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_9346a7c0
We argue that this optimization difficulty is unlikely to be caused by vanishing gradients.

Experiments · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_00908340
In fact, the 34-layer plain net is still able to achieve competitive accuracy (Table 3), suggesting that the solver works to some extent.

Experiments · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_beb44ca7
First, the situation is reversed with residual learning – the 34-layer ResNet is better than the 18-layer ResNet (by 2.8%).

Experiments · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_0f3d3fee
More importantly, the 34-layer ResNet exhibits considerably lower training error and is generalizable to the validation data.

Experiments · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_0c33a7d3
We argue that this is because the zero-padded dimensions in A indeed have no residual learning.

Experiments · page 6 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_d62afd15
Remarkably, although the depth is significantly increased, the 152-layer ResNet (11.3 billion FLOPs) still has lower complexity than VGG-16/19 nets (15.3/19.6 billion FLOPs).

Experiments · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_43a9f1a4
Our baseline 34-layer ResNets have achieved very competitive accuracy.

Experiments · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_7e1194bd
This single-model result outperforms all previous ensemble results (Table 5).

Experiments · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_c630da1e
It has fewer parameters than other deep and thin 5With an initial learning rate of 0.1, it starts converging ( <90% error) after several epochs, but still reaches similar accuracy.

CIFAR-10 and Analysis · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_f3fe2ba3
The error of plain-110 is higher than 60% and not displayed.

CIFAR-10 and Analysis · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_3bef384a
Our method shows no optimization difficulty , and this 103-layer network is able to achieve training error <0.1% (Fig. 6, right).

CIFAR-10 and Analysis · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_43aed781
Its test error is still fairly good (7.93%, Table 6).

CIFAR-10 and Analysis · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_73b8f902
Object detection mAP (%) on the PASCAL VOC 2007/2012 test sets using baseline Faster R-CNN.

CIFAR-10 and Analysis · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_7a376285
See also Table 9 for better results. have similar training error.

CIFAR-10 and Analysis · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_f1d76064
We argue that this is because of overfitting.

CIFAR-10 and Analysis · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_cf2b5755
Here we are interested in the improvements of replacing VGG-16 [41] with ResNet-101.

Object Detection on PASCAL and MS COCO · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_24dca08d
Most remarkably, on the challenging COCO dataset we obtain a 6.0% increase in COCO’s standard metric (mAP@[.5, .95]), which is a 28% relative improvement.

Object Detection on PASCAL and MS COCO · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_8eb755bb
ResNet-101 has a 6% increase of mAP@[.5, .95] over VGG-16, which is a 28% relative improvement, solely contributed by the features learned by the better network.

Object Detection Baselines · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_754f6c20
For completeness, we report the improvements made for the competitions.

Object Detection Improvements · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_91fce645
Box refinement improves mAP by about 2 points (Table 9).

Object Detection Improvements · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_4722bc81
Global context improves mAP@.5 by about 1 point (Table 9).

Object Detection Improvements · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_1f74fc32
Multi-scale testing improves the mAP by over 2 points (Table 9).

Object Detection Improvements · page 11 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_744ed5f5
Our detection system is Faster R-CNN [32] with the improvements in Table 9, using ResNet-101. we achieve 85.6% mAP on PASCAL VOC 2007 (Table 10) and 83.8% on PASCAL VOC … (excerpt)

Object Detection Improvements · page 11 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_23b3f423
58.8% mAP and our ensemble of 3 models has 62.1% mAP on the DET test set (Table 12).This result won the 1st place in the ImageNet detection task in ILSVRC … (excerpt)

method · page 12 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_bed0d1e2
Using an ensemble of networks for both classification and localization, we achieve a top-5 localization error of 9.0% on the test set.

ImageNet Localization · page 12 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_d7de4b17
This number significantly outperforms the ILSVRC 14 results (Table 14), showing a 64% relative reduction of error.

ImageNet Localization · page 12 · Automated extraction, not reviewed · 0 reviewed reproductions

Reproduced one of these claims?

Record your result in the Mearc app on your own machine, then use Publish this result with this paper's id (1512.03385) and the claim id shown next to the claim. Mearc creates a file that holds only what you choose to publish. Add it to the public claim data by pull request; a maintainer other than you reviews it before it counts.