Research map / Detection and segmentation / Open-set and few-shot detection
Learning with Free Object Segments for Long-Tailed Instance Segmentation
Cheng Zhang, Tai-Yu Pan, Tianle Chen, Jike Zhong, Wenjin Fu, Wei-Lun Chao
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 |
|---|---|---|---|
| add free object segments pasted onto LVIS images | MosaicOS, ResNet-50 · LVIS v1 | mask AP: 24.45 -> 25.19 | Improved |
| add free segments on top of calibration | MosaicOS, ResNet-50 · LVIS v1 | APr: 23.86 -> 25.11 | Improved |
| rank segments and keep top images (966K) instead of random 966K (ablation) | free segments · LVIS v1 | mask AP: 24.50 -> 25.19 | Improved |
| rank images and keep 966K instead of using all 1,830K (ablation) | free segments · LVIS v1 | mask AP: 24.87 -> 25.19 | Improved |
| use proposed ranking score instead of IoU (ablation) | free segments · LVIS v1 | mask AP: 24.74 -> 25.19 | Improved |
| add drop-rate filtering of segments (ablation) | free segments · LVIS v1 | mask AP: 24.81 -> 25.19 | Improved |
| paste segments onto LVIS images instead of training on object-centric images (ablation) | free segments · LVIS v1 | mask AP: 24.78 -> 25.19 | Improved |
| add free segments to ground-truth copy-paste | MosaicOS · LVIS v1 | mask AP: 24.57 -> 25.36 | Improved |
| replace GT segment pasting with free segments | Mask R-CNN · LVIS v1 | mask AP: 24.06 -> 24.28 | Improved |
| add second training stage (ablation) | free segments, Mask R-CNN · LVIS v1 | mask AP: 23.35 -> 24.28 | Improved |
| use free segments instead of self-training with LVIS network | MosaicOS · LVIS v1 | mask AP: 24.70 -> 25.19 | Improved |
| train 2x schedule instead of 1x | MosaicOS w/ free segments · LVIS v1 | mask AP: 25.19 -> 26.80 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MosaicOS — MosaicOS: A Simple and Effective Use of Object-Centric Images for Long-Tailed Object Detection (ICCV 2021) · better on 4 of 4 shared results
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
- The Devil is in Classification: A Simple Framework for Long-tail Instance Segmentation (ECCV 2020 · similarity 0.91)
- SOLO: A Simple Framework for Instance Segmentation (TPAMI 2021 · similarity 0.90)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.90)
- Object Affinity Learning: Towards Annotation-Free Instance Segmentation (TPAMI 2023 · similarity 0.90)
- SOLO: Segmenting Objects by Locations (ECCV 2020 · similarity 0.90)
- Training-Free Dataset Pruning for Instance Segmentation (ICLR 2025 · similarity 0.89)
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.
