Research map / Detection and segmentation / Open-set and few-shot detection
Bridging the Gap to Real-World Object-Centric Learning
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 |
|---|---|---|---|
| Reconstruct frozen self-supervised features instead of the input image | Slot Attention object discovery · MOVi-C and MOVi-E | FG-ARI | Helped |
| Reconstruct frozen self-supervised features instead of the input image | Slot Attention object discovery · Real-world datasets | object discovery quality | Helped |
| Add feature reconstruction as the training signal (ablation) | DINOSAUR · COCO and PASCAL VOC | object discovery quality | Helped |
| Replace the MLP decoder with the Transformer decoder (ablation) | DINOSAUR · PASCAL VOC and COCO | class mBO: 9--10 class mBO | Helped |
| Replace the MLP decoder with the Transformer decoder (ablation) | DINOSAUR · PASCAL VOC and COCO | FG-ARI | Hurt |
| Use the MLP decoder instead of the Transformer decoder (ablation) | DINOSAUR · PASCAL VOC and COCO | instance grouping | Helped |
| Use the Transformer decoder instead of the MLP decoder (ablation) | DINOSAUR · PASCAL VOC and COCO | semantic grouping | Helped |
| Train a ResNet34 encoder from scratch rather than use a pretrained ViT encoder (ablation) | DINOSAUR · COCO | object discovery quality | No clear effect |
| Use DINO, MoCo-v3, MSN, or MAE pretrained features as reconstruction targets (ablation) | DINOSAUR with a scratch ResNet34 encoder · COCO | object discovery quality | No clear effect |
| Scale Slot Attention while retaining image reconstruction (ablation) | Slot Attention · Real-world datasets | object grouping quality | No clear effect |
| Use a Transformer decoder rather than the MLP decoder (ablation) | DINOSAUR · MOVi-C | mask quality | Mixed |
Closest papers by meaning
- Object-Centric Pretraining via Target Encoder Bootstrapping (ICLR 2025 · similarity 0.87)
- Temporally Consistent Object-Centric Learning by Contrasting Slots (CVPR 2025 · similarity 0.87)
- A Data-Centric Revisit of Pre-Trained Vision Models for Robot Learning (CVPR 2025 · similarity 0.87)
- HUNTER: Unsupervised Human-centric 3D Detection via Transferring Knowledge from Synthetic Instances to Real Scenes (CVPR 2024 · similarity 0.87)
- Multi-Part Object Representations via Graph Structures and Co-Part Discovery (NeurIPS 2026 · similarity 0.86)
- Object Concepts Emerge from Motion (NeurIPS 2025 · similarity 0.86)
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.