AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Generalized and Discriminative Few-Shot Object Detection via SVD-Dictionary Enhancement

Aming WU, Suqi Zhao, Cheng Deng, Wei Liu

NeurIPS 2021 · Detection and segmentation › Open-set and few-shot detection · approach: Few-shot object detection · 18 citations (OpenAlex)

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 changedWhereMeasuredEffect
add SVD-Dictionary enhancement to MPSRMPSR · COCO 30-shotAP: 13.8 -> 16.2Improved
add SVD-Dictionary enhancement to FSCEFSCE · COCO 30-shotAP: 15.4 -> 16.0Improved
add SVD-Dictionary enhancement to MPSRMPSR · PASCAL VOC Split 1 2-shotnAP50: 41.2 -> 47.4Improved
add SVD-Dictionary enhancement to FSCEFSCE · PASCAL VOC Split 2 10-shotnAP50: 48.7 -> 48.2No clear effect
add SVD-Dictionary enhancement to FSCEFSCE · PASCAL VOC Split 1 3-shotnAP75: 20.1 -> 25.1Improved
add dictionary learning on top of SVD generalization (ablation)MPSR + Ours · PASCAL VOC Split 1 2-shotnAP50: 44.3 -> 47.4Improved
add SVD-based generalization only (no dictionary learning) to MPSR (ablation)MPSR · PASCAL VOC Split 1 2-shotnAP50: 41.2 -> 44.3Improved
vary k from 50% to 10% of singular values (ablation)MPSR + Ours · PASCAL VOC Split 1 1-shotnAP50: 41.5 -> 40.6Hurt
vary k from 50% to 90% of singular values (ablation)MPSR + Ours · PASCAL VOC Split 1 2-shotnAP50: 47.4 -> 41.8Hurt
reduce codewords from 24 to 16 (ablation)MPSR + Ours · PASCAL VOC Split 1 1-shotnAP50: 41.5 -> 39.2Hurt
increase codewords from 24 to 32 (ablation)MPSR + Ours · PASCAL VOC Split 1 10-shotnAP50: 61.2 -> 60.3Mixed
add SVD-Dictionary enhancement to MPSRMPSR · PASCAL VOC Split 1 5-shotBase AP50: 69.4 -> 69.8Improved

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy