AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Perspective-Guided Convolution Networks for Crowd Counting

Zhaoyi Yan, Yuchen Yuan, Wangmeng Zuo, Xiao Tan, Yezhen Wang, Shilei Wen, Errui Ding

ICCV 2019 · Architectures and efficiency › CNN design · approach: Filter design and trained-filter analysis · 227 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 perspective-guided convolution blocks (5) to CSRNet* baselineCSRNet* · ShanghaiTech Part AMAE: 67.5 -> 57.0Improved
add 3 PGC blocks to ResNet-101 backbone (ablation)ResNet-101 · ShanghaiTech Part A / BMAE: 109.6/26.2 -> 89.7/18.6Improved
increase Gaussian filter size K from 1 to 7 (ablation)single PGC block · ShanghaiTech Part AMAE: 67.3 -> 65.8Improved
increase Gaussian filter size K from 7 to 9 or larger (ablation)single PGC block · ShanghaiTech Part AMAE: 65.8 -> 66.3 (K=9), 66.4 (K=11)Hurt
stack 5 PGC blocks instead of 1 (ablation)PGCNet · ShanghaiTech Part A / BMAE: 65.8/9.8 -> 57.0/8.8Improved
stack 6 PGC blocks instead of 5 (ablation)PGCNet · ShanghaiTech Part A / BMAE: 57.0/8.8 -> 58.3/9.0Hurt
pre-train PENet (ablation)PGCNet · UCF_CC_50MAE: 278.6 -> 244.6Improved
pre-train PENet (ablation)PGCNet · Crowd SurveillanceMAE: 10.3 -> 7.2Improved
train PENet end-to-end (Ours B) instead of using fixed estimated perspective (Ours A) (ablation)PGCNet · UCF_CC_50MAE: 259.4 -> 244.6Improved
train PENet end-to-end (Ours B) instead of using fixed estimated perspective (Ours A) (ablation)PGCNet · UCF_CC_50MSE: 317.6 -> 361.2Hurt
train PENet end-to-end (Ours B) instead of Ours A (ablation)PGCNet · Crowd SurveillanceMAE: 7.7 -> 7.2Improved
use ground-truth instead of PENet-estimated perspective map (ablation)PGCNet · ShanghaiTech Part A / B, WorldExpo'10MAE: 58.1/9.0/8.3 -> 57.0/8.8/8.1Improved

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