AI Paper Atlas

Research map / Detection and segmentation / Tracking

Learning Policies for Adaptive Tracking With Deep Feature Cascades

Chen Huang, Simon Lucey, Deva Ramanan

ICCV 2017 · Detection and segmentation › Tracking · approach: Siamese network tracking · 238 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 early stopping policy (EAST) instead of always using the last layer (ablation)EAST · OTB-50AUC (OPE): 0.645 -> 0.638Hurt
add early stopping policy (EAST) instead of always using the last layer (ablation)EAST · OTB-50CPU speed (fps): 3.4 -> 23.2Improved
add learned Q-learning stopping policy instead of threshold heuristic (ablation)EAST · OTB-50AUC (OPE): 0.597 -> 0.638Improved
add learned Q-learning policy instead of threshold heuristic (ablation)EAST · OTB-50CPU speed (fps): 45.7 -> 23.2Hurt
add confidence history to the policy state (ablation)EAST · OTB-50AUC (OPE): 0.621 -> 0.638Improved
add action history to the policy state (ablation)EAST · OTB-50AUC (OPE): 0.610 -> 0.638Improved
add HOG layer to the cascade (with Pixel and deep layers) (ablation)EAST · OTB-50AUC (OPE): 0.631 -> 0.638Improved
add Pixel layer to the cascade (with HOG and deep layers) (ablation)EAST · OTB-50AUC (OPE): 0.628 -> 0.638Improved
add both Pixel and HOG layers before deep layers (ablation)EAST · OTB-50AUC (OPE): 0.619 -> 0.638Improved
add both Pixel and HOG layers before deep layers (ablation)EAST · OTB-50CPU speed (fps): 10.7 -> 23.2Improved
use deep supervision with a single-pass scale prediction instead of SiamFC image pyramidEAST_last · OTB-50AUC (OPE): 0.612 -> 0.645Improved
update the template of lower deep layers C1-C2 during tracking (ablation)SiamFC-style deep layers · OTB-50AUC (OPE)Improved

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