AI Paper Atlas

Research map / Detection and segmentation / CNN features

HarDNet: A Low Memory Traffic Network

Ping Chao, Chao-Yang Kao, Yu-Shan Ruan, Chien-Hsiang Huang, Youn-Long Lin

ICCV 2019 · Detection and segmentation › CNN features · approach: Efficient network training and design · 370 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
replace dense connections with harmonic dense connections (same depth and size)FC-DenseNet-ref100 · CamVidmean IoU: 65.8 -> 67.1Improved
replace dense connections with harmonic dense connectionsFC-DenseNet-ref100 · CamVidCIO (MB): 3337 -> 2076Improved
replace dense connections with harmonic dense connectionsFC-DenseNet-ref100 · CamVidGPU time (s): 15.2 -> 10Improved
use harmonic sparsification instead of SparseNet connectionsFC-SparseNet-ref100 · CamVidGPU time (s): 11.8 -> 10Improved
replace FC-DenseNet103 with FC-HarDNet84FC-DenseNet103 · CamVidGPU time (s): 11.4 -> 6.7Improved
replace FC-DenseNet56 with FC-HarDNet68FC-DenseNet56 · CamVidCIO (MB): 1351 -> 473Improved
replace FC-DenseNet56 with FC-HarDNet68FC-DenseNet56 · CamVidmean IoU: 58.9 -> 62.9Mixed
replace ResNet-50 with HarDNet-68ResNet 50 · ImageNetGPU time (ms): 46.5 -> 32.6Improved
replace ResNet-152 with HarDNet-138sResNet 152 · ImageNetGPU time (ms): 109.7 -> 70.5Improved
replace MobileNetV2 with HarDNet-39DSMobileNetV2 · ImageNetGPU time (ms): 23.7 -> 17.8Improved
use HarDNet for small models versus ResNetResNet 18 · ImageNetGPU time (ms)No clear effect
use inverted transition module (extra max pooling branch) (ablation)HarDNet transition layer · ImageNetCIO: reduces 50% of CIOImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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