Research map / Video / Video segmentation and tracking
MaskRNN: Instance Level Video Object Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add long-term recurrent training (ablation) | MaskRNN · DAVIS-2016 and DAVIS-2017 | IoU (%): 53.90 -> 56.32; 41.57 -> 45.53 | Helped |
| Add the flow stream to the appearance stream (ablation) | MaskRNN segmentation net · DAVIS-2016 and DAVIS-2017 | IoU (%) without finetuning: 54.17 -> 55.87; 41.29 -> 43.33 | Helped |
| Warp the previous mask with optical flow (ablation) | MaskRNN segmentation net · DAVIS-2016 and DAVIS-2017 | IoU (%) without finetuning: 55.87 -> 56.88; 43.33 -> 44.52 | Helped |
| Add bounding-box regression training without applying localization (ablation) | MaskRNN · DAVIS-2016 and DAVIS-2017 | IoU (%) without finetuning: 56.88 -> 52.29; 44.52 -> 38.95 | Hurt |
| Apply localization bounding boxes to restrict the segmentation mask (ablation) | MaskRNN · DAVIS-2016 and DAVIS-2017 | IoU (%) without finetuning: 52.29 -> 53.90; 38.95 -> 41.57 | Helped |
| Finetune on the first frame of the test video (ablation) | MaskRNN · DAVIS-2016 and DAVIS-2017 | IoU (%): 56.32 -> 80.38; 45.53 -> 60.51 | Helped |
| Use MaskRNN instead of OSVOS | Video object segmentation · DAVIS-2016 validation | J Mean IoU (%): 79.8 -> 80.4 | Helped |
| Use MaskRNN instead of the strongest listed baseline | Instance-level video object segmentation · DAVIS-2017 validation and SegTrack v2 | IoU (%): 54.9 -> 60.5; 67.5 -> 72.1 | Helped |
Closest papers by meaning
- Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation (CVPR 2020 · similarity 0.89)
- Mask-Free Video Instance Segmentation (CVPR 2023 · similarity 0.89)
- Weakly Supervised Instance Segmentation for Videos With Temporal Mask Consistency (CVPR 2021 · similarity 0.89)
- Learning Video Object Segmentation From Static Images (CVPR 2017 · similarity 0.88)
- Boundary-Aware Instance Segmentation (CVPR 2017 · similarity 0.87)
- S2D: Sparse-To-Dense Keymask Distillation for Unsupervised Video Instance Segmentation (NeurIPS 2026 · 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.