| RTMDet: An Empirical Study of Designing Real-Time Object Detectors | ✓ Link | 44.6 | 67.4 | 47.8 | 22.2 | | 65.5 | 188
(RTX3090) | RTMDet-Ins-x | 2022-12-14 |
| RTMDet: An Empirical Study of Designing Real-Time Object Detectors | ✓ Link | 43.7 | 66 | 47.0 | 20.8 | 48.0 | 64.8 | 271
(RTX3090) | RTMDet-Ins-l | 2022-12-14 |
| RTMDet: An Empirical Study of Designing Real-Time Object Detectors | ✓ Link | 42.1 | 63.9 | 45.1 | 19.3 | 46.4 | 63.1 | 371
(RTX3090) | RTMDet-Ins-m | 2022-12-14 |
| []() | | 41.4 | | | | | | 10 | RetinaMask | |
| []() | | 40.2 | | | | | | 8 | Predicting Future Instance Segmentation | |
| []() | | 39.6 | | | | | | 12 | UOAIS | |
| Sparse Instance Activation for Real-Time Instance Segmentation | ✓ Link | 37.9 | 59.2 | 40.2 | 15.7 | 39.4 | 56.9 | 40 (2080 Ti) | SparseInst-608 (ResNet-50-vd) | 2022-03-24 |
| Mask-aware IoU for Anchor Assignment in Real-time Instance Segmentation | ✓ Link | 37.7 | 59.4 | 39.9 | 18.1 | 40.8 | 52.5 | 25 (Tesla V100) | maYOLACT-700 (ResNet-50) | 2021-10-19 |
| SOLOv2: Dynamic and Fast Instance Segmentation | ✓ Link | 37.1 | 57.7 | 39.7 | | | | 31.3 | SOLO-512 | 2020-03-23 |
| Sparse Instance Activation for Real-Time Instance Segmentation | ✓ Link | 35.9 | 56.5 | 37.7 | 12.3 | 37.1 | 57.0 | 58.5 (2080 Ti) | SparseInst-448 (ResNet-50-vd) | 2022-03-24 |
| SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation | ✓ Link | 35.4 | 55.6 | 37.6 | 11.2 | 38.3 | 56.8 | 27.0 (Titan Xp) | SipMask++ (ResNet-101, single-scale test) | 2020-07-29 |
| Mask-aware IoU for Anchor Assignment in Real-time Instance Segmentation | ✓ Link | 35.2 | 56.2 | 37.1 | 14.7 | 38.0 | 51.4 | 30 (Tesla V100) | maYOLACT-550 (ResNet-50) | 2021-10-19 |
| BlendMask: Top-Down Meets Bottom-Up for Instance Segmentation | ✓ Link | 35.2 | | | | | | 33.3 | BlendMask-512 (DLA_34) | 2020-01-02 |
| YOLACT++: Better Real-time Instance Segmentation | ✓ Link | 34.6 | 53.8 | 36.9 | 11.9 | 36.8 | 55.1 | 27.3 (Titan Xp) | YOLACT-550++ (ResNet-101-FPN) | 2019-12-03 |
| []() | | 34.6 | | | | | | 5 | Mask R-CNN | |
| CenterMask : Real-Time Anchor-Free Instance Segmentation | ✓ Link | 32.9 | | | 12.9 | 34.7 | 48.7 | | CenterMask-Lite (ResNet-50-FPN) | 2019-11-15 |
| SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation | ✓ Link | 32.8 | 53.4 | 34.3 | 9.3 | 35.6 | 54.0 | 31.3 (Titan Xp) | SipMask (ResNet-101, single-scale test) | 2020-07-29 |
| SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation | ✓ Link | 31.2 | 51.9 | 32.3 | 9.2 | 33.6 | 49.8 | 41.7 (Titan Xp) | SipMask (ResNet-50, single-scale test) | 2020-07-29 |
| []() | | 29.5 | | | | | | 172.7 | YolactEdge | |
| YOLACT: Real-time Instance Segmentation | ✓ Link | 28.2 | 46.6 | 29.2 | 9.2 | 29.3 | 44.8 | 33.3 (Titan Xp) | YOLACT-550 (ResNet-101-FPN) | 2019-04-04 |
| YOLACT: Real-time Instance Segmentation | ✓ Link | 24.9 | 42.0 | 25.4 | 5.0 | 25.3 | 45.0 | 45.3 (Titan Xp) | YOLACT | 2019-04-04 |
| RTMDet: An Empirical Study of Designing Real-Time Object Detectors | ✓ Link | | 59.3 | 41.3 | 15.1 | 42.3 | 60.3 | 518
(RTX3090) | RTMDet-Ins-s | 2022-12-14 |