AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

EdgePoint2: Compact Descriptors for Superior Efficiency and Accuracy

Haodi Yao, Fenghua He, Ning Hao, Xie Chen

IJCV 2026 · Architectures and efficiency › RNNs and normalization · approach: Dataset condensation and flow estimation · 1 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
replace PCA with LRA compression for L_op distillation (ablation)EdgePoint2-M48 · MegaDepth-1500 ablation (EdgePoint2-M48)AUC@5: 50.89 -> 51.75Improved
use Orthogonal Procrustes loss with LRA instead of HF-Net distillation with LRA (ablation)EdgePoint2-M48 · MegaDepth-1500 ablation (EdgePoint2-M48)AUC@5: 48.13 -> 51.75Improved
use Orthogonal Procrustes loss instead of Similarity-Preserving distillation (ablation)EdgePoint2-M48 · MegaDepth-1500 ablation (EdgePoint2-M48)AUC@5: 44.53 -> 51.75Improved
use Orthogonal Procrustes loss with LRA instead of AWDesc distillation (ablation)EdgePoint2-M48 · MegaDepth-1500 ablation (EdgePoint2-M48)AUC@5: 51.18 -> 51.75Improved
add image augmentation (N=4) to the Procrustes loss (ablation)EdgePoint2-M48 · MegaDepth-1500 ablation (EdgePoint2-M48)AUC@5: 51.75 -> 52.17Improved
add similarity loss L_sim (ablation)EdgePoint2-M48 · MegaDepth-1500 ablation (EdgePoint2-M48)AUC@5: 52.17 -> 52.07Mixed
increase model size from T32 to M32EdgePoint2 · MegaDepth-1500AUC@5: 46.12 -> 49.75Improved
increase descriptor dimension from 32 to 48 (M)EdgePoint2 · MegaDepth-1500AUC@5: 49.75 -> 52.07Improved
replace EdgePoint with EdgePoint2-T32 (32-dim)EdgePoint · MegaDepth-1500AUC@5: 42.92 -> 46.12Improved
replace EdgePoint with EdgePoint2-T32EdgePoint · MegaDepth-1500FPS on Orin-NX GPU: 838.8 -> 508.7Hurt

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