res_9cec3e17MoCo can outperform its supervised pre-training counterpart in 7 detection/segmentation tasks on PASCAL VOC, COCO, and other datasets, sometimes surpassing it by large margins.
arXiv 1911.05722 · cs.CV
Read on arXiv · arXiv non-exclusive license
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.
res_9cec3e17MoCo can outperform its supervised pre-training counterpart in 7 detection/segmentation tasks on PASCAL VOC, COCO, and other datasets, sometimes surpassing it by large margins.
res_9d1cd14cWe show that in 7 downstream tasks related to detection or segmentation, MoCo unsupervised pre-training can surpass its ImageNet supervised counterpart, in some cases by nontrivial margins.
clm_a79e8193We propose a momentum update to address this issue.
clm_ea9ad6b6A similar trend has been observed in [61, 56] under the memory bank mechanism, while here we show that this trend is more general and can be seen in all … (excerpt)
res_ac88ac4eBut it is 2.6% worse than MoCo.
res_44ffb4bdNote the memory bank result of 58.0% reflects our improved implementation of [61].2 Ablation: momentum.
res_41059ed4The table below shows ResNet-50 accuracy with different MoCo momentum values ( m in Eqn.(2)) used in pre-training (K = 4096 here) : momentum m 0 0.9 0.99 0.999 0.9999 … (excerpt)
res_ca7015b8MoCo with R50 performs competitively and achieves 60.6% accuracy, better than all competitors of similar model sizes (∼24M).
res_ae888a2fMoCo benefits from larger models and achieves 68.6% accuracy with R50w4×.
res_2dde224dNotably, we achieve competitive results using astandard ResNet-50 and require no specific architecture designs,e.g., 2Here 58.0% is with InfoNCE and K=65536.
res_5d1dc942For R50-dilated- C5 (Table 2a), MoCo pre-trained on IN-1M is comparable to the supervised pre-training counterpart, and MoCo pretrained on IG-1B surpasses it.
clm_02ac533dWe point out that these results are partially because we establish solid detection baselines for contrastive learning.
res_6bac1dccIn green are the gaps of at least +0.5 point. †: this entry is with BN frozen, which improves results; see main text. with ImageNet supervised pre-training: COCO keypoint detection … (excerpt)
res_aa057dc8MoCo with IG-1B surpasses it in all metrics.
res_13dd269bIn sum, MoCo can outperform its ImageNet supervised pre-training counterpart in 7 detection or segmentation tasks.5 Besides, MoCo is on par on Cityscapes instance segmentation, and lags behind on VOC … (excerpt)
res_d57c193dMoCo’s improvement from IN-1M to IG-1B is consistently noticeable but relatively small, suggesting that the larger-scale data may not be fully exploited.
res_ca1d7558MoCo has 24.1 AP mk with IN-1M and 24.9 APmk with IG-1B, both outperforming the supervised pretraining counterpart under the same tunable BN setting.
res_37f7f73bUnder the best individual settings, MoCo can still outperform the supervised pre-training case (24.9 vs. 24.4, as reported in Table 6 in Sec 4.2).
res_ea2a1420IN-1M MoCoIN-1M MoCoIG-1B accuracy (%) 61.8 66.1 65.6 65.8 MoCo is ∼4% better than training from random initialization, and is closely comparable with its ImageNet supervised counterpart.
res_1239acb5But unsupervised pre-training in the separate, unlabeled dataset of IG-1B represents a typical scenario: in this case, MoCo improves by 0.8%. pre-train APbb APbb50 APbb75 APmk APmk50 APmk75 random init.
res_1c2c2662We observe: (i) fine-tuning with ImageNet-supervised pre-training still has improvements (41.9 AP bb); (ii) training from scratch largely catches up (41.4 AP bb); (iii) the MoCo counterparts improve further (e.g., … (excerpt)
Record your result in the Mearc app on your own machine, then use Publish this result with this paper's id (1911.05722) 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.