Papers › SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation
SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation
Jiale Cao, Rao Muhammad Anwer, Hisham Cholakkal, Fahad Shahbaz Khan, Yanwei Pang, Ling Shao
Single-stage instance segmentation approaches have recently gained popularity due to their speed and simplicity, but are still lagging behind in accuracy, compared to two-stage methods. We propose a fast single-stage instance segmentation method, called SipMask, that preserves instance-specific spatial information by separating mask prediction of an instance to different sub-regions of a detected bounding-box. Our main contribution is a novel light-weight spatial preservation (SP) module that generates a separate set of spatial coefficients for each sub-region within a bounding-box, leading to improved mask predictions. It also enables accurate delineation of spatially adjacent instances. Further, we introduce a mask alignment weighting loss and a feature alignment scheme to better correlate mask prediction with object detection. On COCO test-dev, our SipMask outperforms the existing single-stage methods. Compared to the state-of-the-art single-stage TensorMask, SipMask obtains an absolute gain of 1.0% (mask AP), while providing a four-fold speedup. In terms of real-time capabilities, SipMask outperforms YOLACT with an absolute gain of 3.0% (mask AP) under similar settings, while operating at comparable speed on a Titan Xp. We also evaluate our SipMask for real-time video instance segmentation, achieving promising results on YouTube-VIS dataset. The source code is available at https://github.com/JialeCao001/SipMask.
In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.
Code
Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.
Code Syntology ran Syntology
Not run by Syntology. Nothing on this page verifies that the listed code works.
Tasks
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Instance Segmentation | COCO test-dev | SipMask (ResNet-101, single-scale test) | AP50 | 60.2 | #87 of 112 | Archive leaderboard | report |
| Instance Segmentation | COCO test-dev | SipMask (ResNet-101, single-scale test) | AP75 | 40.8 | #87 of 112 | Archive leaderboard | report |
| Instance Segmentation | COCO test-dev | SipMask (ResNet-101, single-scale test) | APL | 54.3 | #87 of 112 | Archive leaderboard | report |
| Instance Segmentation | COCO test-dev | SipMask (ResNet-101, single-scale test) | APM | 40.8 | #87 of 112 | Archive leaderboard | report |
| Instance Segmentation | COCO test-dev | SipMask (ResNet-101, single-scale test) | APS | 17.8 | #87 of 112 | Archive leaderboard | report |
| Instance Segmentation | COCO test-dev | SipMask (ResNet-101, single-scale test) | mask AP | 38.1 | #87 of 112 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask++ (ResNet-101, single-scale test) | AP50 | 55.6 | #11 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask++ (ResNet-101, single-scale test) | AP75 | 37.6 | #11 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask++ (ResNet-101, single-scale test) | APL | 56.8 | #11 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask++ (ResNet-101, single-scale test) | APM | 38.3 | #11 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask++ (ResNet-101, single-scale test) | APS | 11.2 | #11 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask++ (ResNet-101, single-scale test) | Frame (fps) | 27.0 (Titan Xp) | #11 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask++ (ResNet-101, single-scale test) | mask AP | 35.4 | #11 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-101, single-scale test) | AP50 | 53.4 | #17 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-101, single-scale test) | AP75 | 34.3 | #17 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-101, single-scale test) | APL | 54.0 | #17 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-101, single-scale test) | APM | 35.6 | #17 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-101, single-scale test) | APS | 9.3 | #17 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-101, single-scale test) | Frame (fps) | 31.3 (Titan Xp) | #17 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-101, single-scale test) | mask AP | 32.8 | #17 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-50, single-scale test) | AP50 | 51.9 | #18 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-50, single-scale test) | AP75 | 32.3 | #18 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-50, single-scale test) | APL | 49.8 | #18 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-50, single-scale test) | APM | 33.6 | #18 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-50, single-scale test) | APS | 9.2 | #18 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-50, single-scale test) | Frame (fps) | 41.7 (Titan Xp) | #18 of 22 | Archive leaderboard | report |
| Real-time Instance Segmentation | MSCOCO | SipMask (ResNet-50, single-scale test) | mask AP | 31.2 | #18 of 22 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, ms-train, single-scale test) | AP50 | 54.1 | #36 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, ms-train, single-scale test) | AP75 | 35.8 | #36 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, ms-train, single-scale test) | AR1 | 35.4 | #36 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, ms-train, single-scale test) | AR10 | 40.1 | #36 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, ms-train, single-scale test) | mask AP | 33.7 | #36 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, single-scale test) | AP50 | 53 | #38 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, single-scale test) | AP75 | 33.3 | #38 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, single-scale test) | AR1 | 33.5 | #38 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, single-scale test) | AR10 | 38.9 | #38 of 44 | Archive leaderboard | report |
| Video Instance Segmentation | YouTube-VIS validation | SipMask (ResNet-50, single-scale test) | mask AP | 32.5 | #38 of 44 | Archive leaderboard | report |
Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections