Research map / Detection and segmentation / Open-set and few-shot detection
Beyond Max-Margin: Class Margin Equilibrium for Few-Shot Object Detection
Check 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 class margin equilibrium to MPSR | MPSR · MS COCO 10-shot | AP: 9.8 -> 15.1 | Helped |
| Add class margin equilibrium to MPSR | MPSR · MS COCO 30-shot | AP: 14.1 -> 16.9 | Helped |
| Add max-margin loss (ablation) | Meta YOLO · Pascal VOC split-1, 1-shot | detection performance: 14.8 -> 13.5 | Hurt |
| Add max-margin loss (ablation) | Meta YOLO · Pascal VOC split-1, 2-, 3-, and 5-shot | detection performance: 15.5 -> 21.9; 26.7 -> 28.5; 33.9 -> 40.2 | Helped |
| Add feature filtering to max-margin training (ablation) | Meta YOLO with max-margin loss · Pascal VOC split-1, 5-shot | detection performance: 40.2 -> 43.1 | Helped |
| Add gradient-based feature disturbance (ablation) | Meta YOLO with max-margin and feature filtering · Pascal VOC split-1, 1-, 2-, 3-, and 5-shot | detection performance: 13.2 -> 17.8; 23.4 -> 26.1; 29.9 -> 31.5; 43.1 -> 44.8 | Helped |
| Use 1024 output channels for feature filtering (ablation) | Meta YOLO with feature filtering · Pascal VOC split-1 | avg. Δ: -1.2 | Hurt |
| Use 512 output channels for feature filtering (ablation) | Meta YOLO with feature filtering · Pascal VOC split-1 | avg. Δ: +2.9 | Helped |
| Apply feature disturbance to novel classes only (ablation) | Meta YOLO · Pascal VOC split-1 | avg. Δ: -0.2 | Hurt |
| Apply feature disturbance to base classes only (ablation) | Meta YOLO · Pascal VOC split-1 | avg. Δ: +1.7 | Helped |
| Replace gradient truncation with random sampling (ablation) | Meta YOLO feature disturbance · Pascal VOC split-1 | avg. Δ: +0.3 | Helped |
| Replace gradient truncation with random cropping (ablation) | Meta YOLO feature disturbance · Pascal VOC split-1 | avg. Δ: +0.2 | Helped |
Closest papers by meaning
- DiGeo: Discriminative Geometry-Aware Learning for Generalized Few-Shot Object Detection (CVPR 2023 · similarity 0.88)
- Few-Shot Object Detection via Association and DIscrimination (NeurIPS 2021 · similarity 0.88)
- Few-Shot Object Detection via Variational Feature Aggregation (AAAI 2023 · similarity 0.87)
- A Closer Look at Few-shot Classification (ICLR 2019 · similarity 0.87)
- Few-Class Arena: A Benchmark for Efficient Selection of Vision Models and Dataset Difficulty Measurement (ICLR 2025 · similarity 0.85)
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 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.