Research map / Detection and segmentation / Open-set and few-shot detection
Generating Features With Increased Crop-Related Diversity for Few-Shot Object Detection
Jingyi Xu, Hieu Le, Dimitris Samaras
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 vanilla conditional-VAE generated features | DeFRCN · PASCAL VOC Novel Split 1, 1-shot | nAP50: 56.3 -> 60.0 | Improved |
| Control generated crop variation by rescaling latent-code norm according to crop IoU (ablation) | DeFRCN with VAE-generated features · PASCAL VOC Novel Split 1, 1-shot | nAP50: 60.0 -> 62.1 | Improved |
| Add Norm-VAE generated features | DeFRCN · MS COCO novel classes, 1-shot | nAP: 6.6 -> 9.5 | Improved |
| Add crop-diverse generated features | DeFRCN · PASCAL VOC Novel Split 1, 1-shot hard proposals | AP50–75: 16.6 -> 18.8 | Improved |
| Train vanilla VAE on augmented rather than original data (ablation) | Vanilla-VAE feature generation · PASCAL VOC Novel Split 1, 1-shot | nAP50: 60.0 -> 60.1 | No clear effect |
| Use only low-IoU/harder generated features instead of diverse feature norms (ablation) | Norm-VAE training of DeFRCN · PASCAL VOC Novel Split 1, 1-shot | nAP50: 60.2 vs 60.9 | Improved |
| Use only high-IoU/easier generated features instead of diverse feature norms (ablation) | Norm-VAE training of DeFRCN · PASCAL VOC Novel Split 1, 1-shot | nAP75: 30.5 vs 31.6 | Improved |
| Use only one difficulty group of generated features rather than varied norms (ablation) | Norm-VAE training of DeFRCN · PASCAL VOC Novel Split 1 | nAP50 | Hurt |
| Use Word2Vec rather than CLIP semantic embeddings (ablation) | Norm-VAE · PASCAL VOC Novel Split 1, 1-shot | nAP50: 62.1 -> 61.6 | Hurt |
| Replace latent-norm control with β-VAE latent-dimension traversal (ablation) | DeFRCN with generated features · PASCAL VOC Novel Split 1, 1-shot | nAP50: 62.1 -> 61.3 | Hurt |
| Replace latent-norm control with CSVAE latent-subspace manipulation (ablation) | DeFRCN with generated features · PASCAL VOC Novel Split 1, 1-shot | nAP50: 62.1 -> 61.6 | Hurt |
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.
- DeFRCN — DeFRCN: Decoupled Faster R-CNN for Few-Shot Object Detection (ICCV 2021) · better on 5 of 5 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
- Few-Shot Object Detection via Variational Feature Aggregation (AAAI 2023 · similarity 0.86)
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 2024 · similarity 0.86)
- Depth-Aware Concealed Crop Detection in Dense Agricultural Scenes (CVPR 2024 · similarity 0.85)
- Fixing Background Misclassification in Few-Shot Object Detection via Product of Experts (TPAMI 2025 · similarity 0.85)
- Training-free Detection of AI-generated images via Cropping Robustness (NeurIPS 2025 · similarity 0.85)
- Cropper: Vision-Language Model for Image Cropping through In-Context Learning (CVPR 2025 · similarity 0.84)
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.
