Research map / Detection and segmentation / CNN features
NAS-FCOS: Efficient Search for Object Detection Architectures
Ning Wang, Yang Gao, Hao Chen, Peng Wang, Zhi Tian, Chunhua Shen, Yanning Zhang
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 |
|---|---|---|---|
| search both FPN and head (progressive) instead of head only (ablation) | NAS-FCOS (VOC) R-50 · COCO test-dev | AP: 38.7 -> 39.8 | Improved |
| search both FPN and head instead of FPN only (ablation) | NAS-FCOS (VOC) R-50 · COCO test-dev | AP: 38.9 -> 39.8 | Improved |
| use searched space instead of deformable convs in FPN and head (ablation) | NAS-FCOS (VOC) R-50 · COCO test-dev | AP: 38.7 -> 39.8 | Improved |
| use searched FPN instead of deformable-conv FPN (ablation) | NAS-FCOS (VOC) R-50 · COCO test-dev | AP: 38.4 -> 38.9 | Improved |
| replace standard convs in FCOS with deformable convs (ablation) | FCOS R-50 · COCO test-dev | AP: 37.4 -> 38.7 | Improved |
| use negative detection loss as RL reward instead of AP (ablation) | NAS controller · VOC proxy task | AP | Improved |
| use COCO-subset proxy instead of VOC proxy | NAS-FCOS @256 R-50 · COCO test-dev | AP: 39.8 -> 40.3 | Improved |
| use COCO-subset proxy instead of VOC proxy | NAS-FCOS @256 MobileNetV2 · COCO test-dev | AP: 34.7 -> 36.2 | Improved |
| replace FPN-FCOS decoder by searched decoder | NAS-FCOS (COCO) R-50 · COCO test-dev | AP: 37.4 -> 40.3 | Improved |
| transfer searched structure to RetinaNet | NAS-RetinaNet R-50 · COCO | AP: 36.1 -> 39.5 | Improved |
| use 128-channel searched decoder instead of FCOS decoder | NAS-FCOS (VOC) @128 MobileNetV2 · COCO test-dev | AP and FLOPs: 31.2 -> 32.0; 105.4 -> 39.3 GFLOPs | Improved |
| replace FPN-FCOS by searched 128-256 decoder with improvements | NAS-FCOS ResNeXt-101 64x4d · COCO test-dev | AP: 44.7 -> 46.1 | Improved |
Closest papers by meaning
- NAS-FCOS: Fast Neural Architecture Search for Object Detection (CVPR 2020 · similarity 0.98)
- SM-NAS: Structural-to-Modular Neural Architecture Search for Object Detection (AAAI 2020 · similarity 0.92)
- NAS-FPN: Learning Scalable Feature Pyramid Architecture for Object Detection (CVPR 2019 · similarity 0.91)
- SP-NAS: Serial-to-Parallel Backbone Search for Object Detection (CVPR 2020 · similarity 0.91)
- Auto-FPN: Automatic Network Architecture Adaptation for Object Detection Beyond Classification (ICCV 2019 · similarity 0.90)
- G-NAS: Generalizable Neural Architecture Search for Single Domain Generalization Object Detection (AAAI 2024 · similarity 0.87)
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.
