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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace dense connections with harmonic dense connections (same depth and size) | FC-DenseNet-ref100 · CamVid | mean IoU: 65.8 -> 67.1 | Improved |
| replace dense connections with harmonic dense connections | FC-DenseNet-ref100 · CamVid | CIO (MB): 3337 -> 2076 | Improved |
| replace dense connections with harmonic dense connections | FC-DenseNet-ref100 · CamVid | GPU time (s): 15.2 -> 10 | Improved |
| use harmonic sparsification instead of SparseNet connections | FC-SparseNet-ref100 · CamVid | GPU time (s): 11.8 -> 10 | Improved |
| replace FC-DenseNet103 with FC-HarDNet84 | FC-DenseNet103 · CamVid | GPU time (s): 11.4 -> 6.7 | Improved |
| replace FC-DenseNet56 with FC-HarDNet68 | FC-DenseNet56 · CamVid | CIO (MB): 1351 -> 473 | Improved |
| replace FC-DenseNet56 with FC-HarDNet68 | FC-DenseNet56 · CamVid | mean IoU: 58.9 -> 62.9 | Mixed |
| replace ResNet-50 with HarDNet-68 | ResNet 50 · ImageNet | GPU time (ms): 46.5 -> 32.6 | Improved |
| replace ResNet-152 with HarDNet-138s | ResNet 152 · ImageNet | GPU time (ms): 109.7 -> 70.5 | Improved |
| replace MobileNetV2 with HarDNet-39DS | MobileNetV2 · ImageNet | GPU time (ms): 23.7 -> 17.8 | Improved |
| use HarDNet for small models versus ResNet | ResNet 18 · ImageNet | GPU time (ms) | No clear effect |
| use inverted transition module (extra max pooling branch) (ablation) | HarDNet transition layer · ImageNet | CIO: reduces 50% of CIO | Improved |
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.
- MobileNetV2 — MobileNetV2: Inverted Residuals and Linear Bottlenecks (CVPR 2018) · better on 1 of 1 shared result
- SegNet — SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation (TPAMI 2017) · better on 1 of 1 shared result
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
- Run, Don't Walk: Chasing Higher FLOPS for Faster Neural Networks (CVPR 2023 · similarity 0.87)
- Designing Extremely Memory-Efficient CNNs for On-device Vision and Audio Tasks (IJCV 2026 · similarity 0.86)
- Latency-Aware Unified Dynamic Networks for Efficient Image Recognition (TPAMI 2024 · similarity 0.86)
- DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs (ECCV 2024 · similarity 0.86)
- Latency-aware Spatial-wise Dynamic Networks (NeurIPS 2022 · similarity 0.86)
- Lightweight and Progressively-Scalable Networks for Semantic Segmentation (IJCV 2023 · similarity 0.85)
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.
