AI Paper Atlas

Research map / Detection and segmentation / CNN features

Feature Space Transfer for Data Augmentation

Bo Liu, Xudong Wang, Mandar Dixit, Roland Kwitt, Nuno Vasconcelos

CVPR 2018 · Detection and segmentation › CNN features · approach: Local features and context in CNNs · 104 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
Augment one-shot training features with pose- and depth-conditioned FATTEN features instead of using only the single-instance baselineLinear SVM · SUN-RGBD T1 one-shotaccuracy: 33.74 -> 44.99Improved
Augment one-shot training features with pose- and depth-conditioned FATTEN features instead of using only the single-instance baselineLinear SVM · SUN-RGBD T2 one-shotaccuracy: 23.76 -> 34.70Improved
Augment one-shot training features with pose- and depth-conditioned FATTEN features instead of using only the single-instance baselineLinear SVM · SUN-RGBD T3 one-shotaccuracy: 22.84 -> 32.20Improved
Use FATTEN feature augmentation rather than AGA augmentationOne-shot recognition · SUN-RGBD T1accuracy: 39.10 -> 44.99Improved
Use FATTEN feature augmentation rather than AGA augmentationOne-shot recognition · SUN-RGBD T2accuracy: 30.12 -> 34.70Improved
Use FATTEN feature augmentation rather than AGA augmentationOne-shot recognition · SUN-RGBD T3accuracy: 26.67 -> 32.20Improved
Augment five-shot training features with FATTEN pose- and depth-conditioned featuresLinear SVM · SUN-RGBD T1 five-shotaccuracy: 50.03 -> 58.82Improved
Augment five-shot training features with FATTEN pose- and depth-conditioned featuresLinear SVM · SUN-RGBD T2 five-shotaccuracy: 36.76 -> 50.69Improved
Augment five-shot training features with FATTEN pose- and depth-conditioned featuresLinear SVM · SUN-RGBD T3 five-shotaccuracy: 37.37 -> 47.07Improved
Generate features for varied poses rather than retrieve using real featuresModelNet retrieval · ModelNet testpose retrieval mAP: 54.58 -> 77.62Improved
Generate features for varied poses rather than retrieve using real featuresModelNet retrieval · ModelNet testcategory retrieval mAP: 32.71 -> 28.89Hurt
Generate features for varied poses rather than retrieve using real featuresModelNet retrieval · ModelNet testjoint pose and category retrieval mAP: 23.65 -> 11.07Hurt

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