AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Person Re-identification via Recurrent Feature Aggregation

Yichao Yan, Bingbing Ni, Zhichao Song, Chao Ma, Yan Yan, Xiaokang Yang

ECCV 2016 · Label-efficient and robust learning › Recognition and long-tail · approach: Supervised person re-identification · 250 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 LSTM recurrent aggregation (last node instead of first node features) (ablation)RFA-Net · iLIDS-VID and PRID 2011rank-1 matching rate: about 10% gainImproved
use RFA-Net+RSVM instead of Color&LBP+DVRColor&LBP features · iLIDS-VIDrank-1 matching rate (%): 34.5 -> 49.3Improved
use RFA-Net+RSVM instead of Color&LBP+DVRColor&LBP features · PRID 2011rank-1 matching rate (%): 37.6 -> 58.2Improved
use RFA-Net+RSVM instead of averaged-pool Color&LBP+RSVMColor&LBP features · iLIDS-VIDrank-1 matching rate (%): 23.2 -> 49.3Improved
add RankSVM metric learning over cosine distance (ablation)RFA-Net · iLIDS-VIDrank-1 matching rate (%): 44.5 -> 49.3Improved
add RankSVM metric learning over cosine distance (ablation)RFA-Net · PRID 2011rank-1 matching rate (%): 54.9 -> 58.2Improved
add RankSVM over cosine distance (ablation)RFA-Net · iLIDS-VID rank-20rank-20 matching rate (%): 90.1 -> 90.0No clear effect
add 50% noise frames to test sequencesRFA-Net+RSVM · iLIDS-VIDrank-1 matching rate (%): 49.3 -> 29.8Hurt
add 50% noise frames to test sequencesRFA-Net+RSVM · PRID 2011rank-1 matching rate (%): 58.2 -> 44.7Hurt
average features over 10 test subsequences instead of 1 (ablation)RFA-Net+RSVM · iLIDS-VIDrank-1 matching rate (%): 33.1 -> 49.3Improved
average features over 10 test subsequences instead of 1 (ablation)RFA-Net+RSVM · PRID 2011rank-1 matching rate (%): 45.6 -> 58.2Improved
average 15 test subsequences instead of 10 (ablation)RFA-Net+RSVM · iLIDS-VIDrank-1 matching rate (%): 49.3 -> 47.8No clear effect

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