Research map / Detection and segmentation / CNN features
Real-Time Instance Segmentation With Discriminative Orientation Maps
Wentao Du, Zhiyu Xiang, Shuya Chen, Chengyu Qiao, Yiman Chen, Tingming Bai
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace grid centers with box centroids as base positions (ablation) | OrienMask · COCO val2017 | mask AP: 32.5 -> 33.3 | Improved |
| use larger anchors inherited from YOLOv4 (ablation) | OrienMask · COCO val2017 | mask AP: 33.3 -> 33.8 | Improved |
| merge multi-scale pyramid features for P2 (fpn plus) (ablation) | OrienMask · COCO val2017 | mask AP: 33.8 -> 34.1 | Improved |
| expand valid training area ratio from 1.0 to 1.2 (ablation) | OrienMask · COCO val2017 | mask AP: 32.0 -> 32.5 | Improved |
| expand valid training area ratio from 1.2 to 1.6 (ablation) | OrienMask · COCO val2017 | mask AP: 32.5 -> 31.9 | Hurt |
| increase orientation loss weight lambda from 5 to 20 (ablation) | OrienMask · COCO val2017 | mask AP: 31.0 -> 32.5 | Improved |
| increase orientation loss weight lambda (ablation) | OrienMask · COCO val2017 | box AP: 36.7 -> 37.4 | No clear effect |
| contraction ratio tau 0.6 instead of 0.4 (ablation) | OrienMask · COCO val2017 | mask AP: 30.4 -> 32.5 | Improved |
| contraction ratio tau 0.6 instead of 0.8 (ablation) | OrienMask · COCO val2017 | mask AP: 30.1 -> 32.5 | Improved |
| use OrienMask instead of YOLACT | real-time instance segmentation · COCO val2017 | mask AP: 28.9 -> 34.5 | Improved |
| use OrienMask instead of YOLACT | real-time instance segmentation · COCO val2017 | FPS: 45.9 -> 41.9 | Hurt |
| use OrienMask instead of BlendMask for mask construction | real-time instance segmentation · COCO val2017 | mask feature memory (MB): 18.44M -> 1.27M | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 4 on most shared results.
- BlendMask — BlendMask: Top-Down Meets Bottom-Up for Instance Segmentation (CVPR 2020) · better on 2 of 4 shared results
- CenterMask — CenterMask: Real-Time Anchor-Free Instance Segmentation (CVPR 2020) · better on 2 of 2 shared results
- PolarMask — PolarMask: Single Shot Instance Segmentation With Polar Representation (CVPR 2020) · better on 1 of 1 shared result
- SOLO — SOLO: Segmenting Objects by Locations (ECCV 2020) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation (ECCV 2020 · similarity 0.90)
- YOLACT: Real-Time Instance Segmentation (ICCV 2019 · similarity 0.89)
- YOLACT++ Better Real-Time Instance Segmentation (TPAMI 2020 · similarity 0.89)
- DynaMask: Dynamic Mask Selection for Instance Segmentation (CVPR 2023 · similarity 0.89)
- BlendMask: Top-Down Meets Bottom-Up for Instance Segmentation (CVPR 2020 · similarity 0.88)
- Any3DIS: Class-Agnostic 3D Instance Segmentation by 2D Mask Tracking (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
