Research map / Detection and segmentation / CNN features
Learning Efficient Single-stage Pedestrian Detectors by Asymptotic Localization Fitting
Wei Liu, Shengcai Liao, Weidong Hu, Xuezhi Liang, Xiao Chen
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 |
|---|---|---|---|
| add second ALF step (ablation) | ALFNet-1s · CityPersons val (IoU=0.5) | MR-2: 16.01 -> 12.01 | Improved |
| add second ALF step (ablation) | ALFNet-1s · CityPersons val (IoU=0.75) | MR-2: 48.95 -> 36.49 | Improved |
| add third ALF step (ablation) | ALFNet-2s · CityPersons val (IoU=0.5) | MR-2: 12.01 -> 12.88 | No clear effect |
| add third ALF step (ablation) | ALFNet-2s · CityPersons val | F-measure: 0.717 -> 0.780 | Improved |
| fuse step scores by product instead of using step 1 scores (ablation) | ALFNet-2s · CityPersons val (IoU=0.5) | MR-2: 13.17 -> 12.01 | Improved |
| fuse step scores instead of using step 2 scores alone (ablation) | ALFNet-2s · CityPersons val (IoU=0.75) | MR-2: 34.90 -> 36.49 | Hurt |
| use step 2 predictor instead of step 1 predictor (ablation) | ALFNet-2s · CityPersons val (IoU=0.75) | MR-2: 46.83 -> 34.90 | Improved |
| raise step 2 IoU thresholds from {0.3,0.5} to {0.5,0.7} (ablation) | ALFNet-2s · CityPersons val (IoU=0.5) | MR-2: 13.75 -> 12.01 | Improved |
| use same high thresholds {0.5,0.7} in both steps instead of increasing (ablation) | ALFNet-2s · CityPersons val (IoU=0.5) | MR-2: 12.01 -> 13.72 | Hurt |
| add ALF module to ResNet-50 (ablation) | ALFNet · CityPersons val (IoU=0.5) | MR-2: 16.01 -> 12.01 | Improved |
| add ALF module to MobileNet (ablation) | ALFNet · CityPersons val (IoU=0.5) | MR-2: 18.88 -> 15.45 | Improved |
| initialize from CityPersons-trained model | ALFNet · Caltech test (IoU=0.5) | MR-2: 6.1 -> 4.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Faster-RCNN — Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016) · 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
- Accurate Single Stage Detector Using Recurrent Rolling Convolution (CVPR 2017 · similarity 0.87)
- Improving Occlusion and Hard Negative Handling for Single-Stage Pedestrian Detectors (CVPR 2018 · similarity 0.86)
- How Far Are We From Solving Pedestrian Detection? (CVPR 2016 · similarity 0.86)
- Is Faster R-CNN Doing Well for Pedestrian Detection? (ECCV 2016 · similarity 0.86)
- Generalizable Pedestrian Detection: The Elephant in the Room (CVPR 2021 · similarity 0.86)
- Learning Rich Features at High-Speed for Single-Shot Object Detection (ICCV 2019 · similarity 0.86)
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.
