Research map / Video / Video segmentation and tracking
STEP: Segmenting and Tracking Every Pixel
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 |
|---|---|---|---|
| Use optical-flow mask propagation for tracking instead of consecutive-frame IoU association | Panoptic-DeepLab baseline · KITTI-STEP | STQ: 0.58 -> 0.67 | Helped |
| Use optical-flow mask propagation for tracking instead of consecutive-frame IoU association | Panoptic-DeepLab baseline · KITTI-STEP | AQ: 0.47 -> 0.63 | Helped |
| Use optical-flow mask propagation for tracking instead of consecutive-frame IoU association | Panoptic-DeepLab baseline · MOTChallenge-STEP | STQ: 0.42 -> 0.41 | No clear effect |
| Use optical-flow mask propagation for tracking instead of consecutive-frame IoU association | Panoptic-DeepLab baseline · MOTChallenge-STEP | AQ: 0.25 -> 0.25 | No clear effect |
| Use unified multi-frame motion prediction instead of separate segmentation and tracking | Motion-DeepLab versus B1 · KITTI-STEP | STQ: 0.58 -> 0.58 | Hurt |
| Use unified multi-frame motion prediction instead of separate segmentation and tracking | Motion-DeepLab versus B1 · MOTChallenge-STEP | STQ: 0.42 -> 0.35 | Hurt |
| Pretrain on Cityscapes before fine-tuning (ablation) | B3: Mask Propagation · KITTI-STEP | STQ: +10% | Helped |
| Pretrain on Cityscapes before fine-tuning (ablation) | B4: Motion-DeepLab · KITTI-STEP | STQ: +4% | Helped |
| Pretrain on Cityscapes before fine-tuning (ablation) | B4: Motion-DeepLab · KITTI-STEP | AQ | Hurt |
| Use appearance-based re-identification cues | VPSNet versus B1 · KITTI-STEP | AQ: 0.47 -> 0.52 | Helped |
| Use appearance-based re-identification cues | VPSNet versus B1 · MOTChallenge-STEP | AQ: 0.25 -> 0.17 | Hurt |
Closest papers by meaning
- Video K-Net: A Simple, Strong, and Unified Baseline for Video Segmentation (CVPR 2022 · similarity 0.89)
- TrackIME: Enhanced Video Point Tracking via Instance Motion Estimation (NeurIPS 2024 · similarity 0.88)
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.88)
- Segment Any Motion in Videos (CVPR 2025 · similarity 0.87)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 2025 · similarity 0.87)
- SMITE: Segment Me In TimE (ICLR 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.