AI Paper Atlas

Research map / Detection and segmentation / CNN features

Improving Multispectral Pedestrian Detection by Addressing Modality Imbalance Problems

Kailai Zhou, Linsen Chen, Xun Cao

ECCV 2020 · Detection and segmentation › CNN features · approach: RGB-D salient object detection · 372 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 DMAF module (ablation)two-stream SSD baseline · KAIST reasonable (all-day)MR (%): 11.93 -> 10.96Improved
add IAFC cascade stage (ablation)two-stream SSD baseline · KAIST reasonable (all-day)MR (%): 11.93 -> 10.53Improved
add DMAF to IAFC model (ablation)IAFC · KAIST reasonable (all-day)MR (%): 10.53 -> 9.36Improved
add modality alignment (ablation)IAFC+DMAF · KAIST reasonable (all-day)MR (%): 9.36 -> 8.13Improved
add modality alignment (ablation)IAFC+DMAF · KAIST reasonable (night)MR (%): 8.63 -> 7.86Improved
add DMAF to IAFC model (ablation)IAFC · KAIST far subsetMR (%): 58.47 -> 54.66Mixed
add illumination gate to RGB+thermal IAFC (ablation)IAFC · KAIST reasonable (all-day)MR (%): 10.27 -> 9.36Improved
use both RGB and thermal IAFC instead of RGB only (ablation)IAFC · KAIST reasonable (all-day)MR (%): 9.86 -> 9.36Improved
use both RGB and thermal IAFC instead of thermal only (ablation)IAFC · KAIST reasonable (all-day)MR (%): 9.89 -> 9.36Improved
add IAFC to DMAF model (ablation)DMAF · KAIST reasonable (all-day)MR (%): 10.96 -> 9.36Improved
MBNet versus AR-CNNMBNet · KAIST reasonable IoU 0.5 (all-day)MR^-2 (%): 9.34 -> 8.13Improved
MBNet versus AR-CNN at stricter IoUMBNet · KAIST reasonable IoU 0.75 (all-day)MR^-2 (%): 64.22 -> 60.12Improved

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