Research map / Detection and segmentation / Open-set and few-shot detection
Spatial Mixture Models with Learnable Deep Priors for Perceptual Grouping
Jinyang Yuan, Bin Li, Xiangyang Xue
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 |
|---|---|---|---|
| use learnable deep priors with disentangled shape weights, appearance components, and a complementary background component | Perceptual grouping on Multi-Shapes · Multi-Shapes 28×28, 3 objects | AMI: 0.875 -> 0.941 | Improved |
| use learnable deep priors with disentangled shape weights, appearance components, and a complementary background component | Perceptual grouping on Multi-Shapes · Multi-Shapes 28×28, 3 objects | MSE: 1.10 × 10^{-2} -> 0.28e-2 | Improved |
| replace truncated stick-breaking weights with softmax weights (ablation) | Proposed method · Multi-Shapes 28×28, 3 objects | AMI: 0.941 -> 0.875 | Hurt |
| replace truncated stick-breaking weights with softmax weights (ablation) | Proposed method · Multi-Shapes 28×28, 3 objects | MSE: 0.28e-2 -> 1.10 × 10^{-2} | Hurt |
| train on 3-object scenes and evaluate on scenes with fewer objects | Proposed method · Multi-Shapes, 2 objects | AMI: 0.906 -> 0.961 | Improved |
| train on 3-object scenes and evaluate on scenes with more objects | Proposed method · Multi-Shapes, 4 objects | AMI: 0.838 -> 0.901 | Improved |
| use Gaussian rather than Bernoulli mixture components (ablation) | Proposed method · Multi-Shapes 28×28, 3 objects, F=64, N=64 | AMI: 0.941 -> 0.939 | No clear effect |
| use Gaussian rather than Bernoulli mixture components (ablation) | Proposed method · Multi-Shapes 28×28, 3 objects, F=64, N=64 | MSE: 0.28e-2 -> 0.29e-2 | No clear effect |
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.
- Tagger — Tagger: Deep Unsupervised Perceptual Grouping (NeurIPS 2016) · better on 3 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
- Deep Grouping Model for Unified Perceptual Parsing (CVPR 2020 · similarity 0.87)
- Disentangling neural mechanisms for perceptual grouping (ICLR 2020 · similarity 0.87)
- Latent Noise Segmentation: How Neural Noise Leads to the Emergence of Segmentation and Grouping (ICML 2024 · similarity 0.87)
- Perceptual Group Tokenizer: Building Perception with Iterative Grouping (ICLR 2024 · similarity 0.86)
- Vision Transformers Learn Gestalt-Like Figure-Ground Cues from Natural Images (NeurIPS 2026 · similarity 0.85)
- Deciphering 'What' and 'Where' Visual Pathways from Spectral Clustering of Layer-Distributed Neural Representations (CVPR 2024 · similarity 0.85)
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.
