MearcAll papers

arXiv 1506.01497 · cs.CV

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Shaoqing Ren, Kaiming He, Ross Girshick, 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 recordedclm_66dedeaf
In this paper, we show that an algorithmic change— computing proposals with a deep convolutional neural network—leads to an elegant and effective solution where proposal computation is nearly cost-free given … (excerpt)

I NTRODUCTION · page 1 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_79948a25
To unify RPNs with Fast R-CNN [2] object detection networks, we propose a training scheme that alternates between fine-tuning for the region proposal task and then fine-tuning for object detection, … (excerpt)

I NTRODUCTION · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_bee3bb8d
1 We comprehensively evaluate our method on the PASCAL VOC detection benchmarks [11] where RPNs with Fast R-CNNs produce detection accuracy better than the strong baseline of Selective Search with … (excerpt)

I NTRODUCTION · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_699c2515
We also report results on the MS COCO dataset [12] and investigate the improvements on PASCAL VOC using the COCO data.

I NTRODUCTION · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_c0b9acfa
[10], we have also found that RPNs can be trained jointly with Fast R-CNN networks leading to less training time. mercial systems such as at Pinterests [17], with user engagement … (excerpt)

Since the publication of the conference version of this paper · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_8c21be53
Multi-Scale Anchors as Regression References Our design of anchors presents a novel scheme for addressing multiple scales (and aspect ratios).

For simplicity we implement the cls layer as a two-class · page 4 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_1a6e12ca
Nevertheless, our method achieves bounding-box regression by a different manner from previous RoIbased (Region of Interest) methods [1], [2].

Loss Function · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_d268742c
RPN with Fast R-CNN achieves competitive results, with an mAP of 59.9% while using up to 300 proposals 8.

Experiments on P ASCAL VOC · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_f43bf9a7
Using RPN yields a much faster detection system than using either SS or EB because of shared convolutional computations; the fewer proposals also reduce the region-wise fully-connected layers’ cost (Table … (excerpt)

Experiments on P ASCAL VOC · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_656ce84b
Somewhat surprisingly, the RPN still leads to a competitive result (55.1%) when using the top-ranked 8.

Experiments on P ASCAL VOC · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_3cf155e3
The mAP is nearly unchanged with N = 1000 (55.8%), but degrades considerably to 44.6% when N = 100 .

Experiments on P ASCAL VOC · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_15c91bbf
The mAP improves from 56.8% (using RPN+ZF) to 59.2% (using RPN+VGG).

Experiments on P ASCAL VOC · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_9c649732
Because proposals of RPN+ZF are competitive with SS (both are 58.7% when consistently used for training and testing), we may expect RPN+VGG to be better than SS.

Experiments on P ASCAL VOC · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_f1effb25
Using RPN+VGG, the result is 68.5% for unshared features, slightly higher than the SS baseline.

Experiments on P ASCAL VOC · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_e89c7ddc
For the feature- shared variant, the result is 69.9%—better than the strong SS baseline, yet with nearly cost-free proposals.

Experiments on P ASCAL VOC · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_55ed0fac
The mAP is higher if using 3 scales (with 1 aspect ratio) or 3 aspect ratios (with 1 scale), demonstrating that using anchors of multiple sizes as the regression references … (excerpt)

Experiments on P ASCAL VOC · page 9 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_0dbcacf7
Using just 3 scales with 1 aspect ratio (69.8%) is as good as using 3 scales with 3 aspect ratios on this dataset, suggesting that scales and aspect ratios are … (excerpt)

Experiments on P ASCAL VOC · page 9 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_b210397d
This explains why the RPN has a good ultimate detection mAP when using as few as 300 proposals.

Experiments on P ASCAL VOC · page 9 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_0d331a32
This is lower than the two-stage system (58.7%) by 4.8%.

Experiments on P ASCAL VOC · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_ea4f8aeb
Including these [0, 0.1) samples improves mAP@0.5 on the COCO dataset for both Fast R-CNN and Faster R-CNN systems (but the impact is negligible on PASCAL VOC).

Experiments on MS COCO · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_230de1a2
Our Fast R-CNN baseline has 39.3% mAP@0.5 on the test-dev set, higher than that reported in [2].

Experiments on MS COCO · page 11 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_c99e7743
Using the COCO training set to train, Faster R-CNN has 42.1% mAP@0.5 and 21.5% mAP@[.5, .95] on the COCO test-dev set.

Experiments on MS COCO · page 11 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_9d802d51
This is 2.8% higher for mAP@0.5 and 2.2% higher for mAP@[.5, .95] than the Fast R- CNN counterpart under the same protocol (Table 11).

Experiments on MS COCO · page 11 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_05ec7cea
Using the COCO trainval set to train, Faster R- CNN has 42.7% mAP@0.5 and 21.9% mAP@[.5, .95] on the COCO test-dev set.

Experiments on MS COCO · page 11 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_6665ddff
This result is better than that trained on VOC07+12 (73.2%) by a good margin, even though the PASCAL VOC data are not exploited.

From MS COCO to P ASCAL VOC · page 11 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_f3f2ecbc
Similar improvements are observed on the PASCAL VOC 2012 test set (Table 12 and Table 7).

From MS COCO to P ASCAL VOC · page 11 · 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 (1506.01497) 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.