clm_4efbfd19Without bells and whistles, Mask R-CNN outperforms all existing, single-model entries on every task, including the COCO 2016 challenge winners.
arXiv 1703.06870 · 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.
clm_4efbfd19Without bells and whistles, Mask R-CNN outperforms all existing, single-model entries on every task, including the COCO 2016 challenge winners.
clm_37ede54bHowever, we show that a surprisingly simple, flexible, and fast system can surpass prior state-of-the-art instance segmentation results.
clm_71b84ce5To fix the misalignment, we propose a simple, quantization-free layer, called RoIAlign, that faithfully preserves exact spatial locations.
res_c6315b12Masks are shown in color, and bounding box, category, and confidences are also shown. a seemingly minor change, RoIAlign has a large impact: it improves mask accuracy by relative 10% … (excerpt)
res_2a2e12bcMask R-CNN surpasses the winner of the 2016 COCO keypoint competition, and at the same time runs at 5 fps.
clm_761e996bNext, we introduce the key elements of Mask R-CNN, including pixel-to-pixel alignment, which is the main missing piece of Fast/Faster R-CNN.
clm_cb1ac55bWe show by experiments that this formulation is key for good instance segmentation results.
clm_6ae587c2To address this, we propose an RoIAlign layer that removes the harsh quantization of RoIPool, properly aligning the extracted features with the input.
res_7417a302RoIAlign leads to large improvements as we show in §4.2.
res_3e6c07d7Note that since we only compute masks on the top 100 detection boxes, Mask R-CNN adds a small overhead to its Faster R-CNN counterpart (e.g., ∼20% on typical models).
clm_c9cdbb1aWe perform a thorough comparison of Mask R-CNN to the state of the art along with comprehensive ablations on the COCO dataset [28].
clm_eb9776f7All instantiations of our model outperform baseline variants of previous state-of-the-art models.
res_9518aeaaOur RoIAlign layer improves AP by ∼3 points and AP75 by∼5 points.
res_12a36433RoIAlign improves AP by about 3 points over RoIPool, with much of the gain coming at high IoU (AP 75).
res_3da3109aAs can be seen in Table 2c, RoIWarp performs on par with RoIPool and much worse than RoIAlign.
res_5176e4c8Mask R-CNN using ResNet-101- FPN outperforms the base variants of all previous state-of-the-art models (the mask output is ignored in these experiments).
res_233457b8As a further comparison, we trained a version of Mask R-CNN but without the mask branch, denoted by “Faster R-CNN, RoIAlign” in Table 3.
res_49d93b9dOn the other hand, it is 0.9 points box AP lower than Mask R-CNN.
res_068b0f85This model runs at 195ms per image on an Nvidia Tesla M40 GPU (plus 15ms CPU time resizing the outputs to the original resolution), and achieves statistically the same mask … (excerpt)
res_3a9a20d7Although Mask R-CNN is fast, we note that our design is not optimized for speed, and better speed/accuracy tradeoffs could be achieved [21],e.g., by varying image sizes and proposal numbers, … (excerpt)
res_522c626bAdding a segment branch (for the person category) improves the AP kp to 63.1 (Table 4) on test-dev.
res_27d4bd9eThough this ResNet-50-FPN backbone has finer strides (e.g., 4 pixels on the finest level), RoIAlign still shows significant improvement over RoIPool and increases APkp by 4.4 points.
res_16c5e90dWithout using the coarse training set, our method achieves 26.2 AP on test, which is over 30% relative improvement over the previous best entry (DIN [3]), and is also better … (excerpt)
res_2d7e0608Compared to the best entry using fine data only (17.4 AP), we achieve a ∼50% improvement.
clm_a54e4f80We argue that within-category overlap is a core difficulty of instance segmentation.
res_be5e01e5Our method shows massive improvement on these two categories over the other best entries (relative ∼40% improvement on person from 21.8 to 30.5 and ∼20% improvement on car from 39.4 … (excerpt)
clm_255ecd8aWe show that using COCO pretraining is an effective strategy on this dataset.
res_2009dd13COCO pre-training helps to improve results the most on these categories; however, the domain shift persists with 38.0/30.1, 57.5/40.9, and 41.2/30.9 val/test AP, respectively.
clm_1a16157fThanks to its generality and flexibility, Mask R-CNN was used as the framework by the three winning teams in the COCO 2017 instance segmentation competition, which all significantly outperformed the … (excerpt)
res_b19ab567Overall, the improvements increase mask AP 5.1 points (from 36.7 to 41.8) and box AP 7.7 points (from 39.6 to 47.3).
res_6a157067Table 8 shows that e2e training improves mask AP by 0.6 and box AP by 1.2.
res_9a89bce3This 5× increase in pre-training data improves both mask and box 1 AP.
res_166717e1Train-time augmentation improves mask AP by 0.6 and box AP by 0.8.
res_cbd8f0f6Using the recently proposed non-local (NL) model [43], we achieve 40.3 mask AP and 45.0 box AP.
res_f5354a32This improves AP kp by about 1 point.
res_e508bb6aWith data distillation, Mask R-CNN AP kp improve by 1.8 points to 69.1.
Record your result in the Mearc app on your own machine, then use Publish this result with this paper's id (1703.06870) 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.