AI Paper Atlas

Research map / Video / Video segmentation and tracking

Jointly learning visual motion and confidence from local patches in event cameras

Daniel R. Kepple, Daewon Lee, Colin Prepsius, Volkan Isler, Il Memming Park, Daniel D. Lee

ECCV 2020 · Video › Video segmentation and tracking · approach: Event-based video reconstruction · 16 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 the learned confidence network (instead of heuristic event-count confidence) (ablation)Ours · DVS-COCO testGlobal MSE (deg/s)^2: 47.1 -> 4.5Improved
add the learned confidence network (instead of heuristic event-count confidence) (ablation)Ours · DVS-COCO testLocal MSE (deg/s)^2: 48.52 -> 6.1Improved
add the local loss Lf on top of the global loss Lg (Mixture of Inputs) (ablation)Ours · DVS-COCO testLocal MSE (deg/s)^2: 54.05 -> 6.1Improved
add the local loss Lf on top of the global loss Lg (Mixture of Inputs) (ablation)Ours · DVS-COCO testGlobal MSE (deg/s)^2: 11.6 -> 4.5Improved
add a second time constant (tau_fast and tau_slow instead of a single tau) (ablation)Ours · DVS-COCO testMSE (deg/s)^2: about 700 -> 4.5Improved
use local network predictions weighted by our confidence versus global contrast maximizationOurs · DVS-COCO testGlobal MSE (deg/s)^2: 6.0 -> 4.5Improved
replace our local flow predictions by local contrast maximization weighted by our confidenceContrast Maximization (local) · DVS-COCO testLocal MSE (deg/s)^2: 6.1 vs 193.7Hurt
weight local CM predictions by variance instead of meanContrast Maximization (local) · DVS-COCO testGlobal MSE (deg/s)^2: 14.9 -> 10.4Improved
use our network confidence weights for local CM instead of variance weightsContrast Maximization (local) · DVS-COCO testGlobal MSE (deg/s)^2: 10.4 -> 28.6Hurt
our method versus Stoffregen on multiple objects sequenceOurs · EED multiple objectssuccess rate (%): 96.77 vs 93.3Hurt
our method versus Stoffregen on lighting variation sequenceOurs · EED lighting variationsuccess rate (%): 80.51 -> 97.40Improved
our network versus EV-FlowNet at 240x180 resolutionOurs · inference latency 240x180latency (ms): 65ms -> 12msImproved

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