AI Paper Atlas

Research map / Detection and segmentation / CNN features

SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization

Xianzhi Du, Tsung-Yi Lin, Pengchong Jin, Golnaz Ghiasi, Mingxing Tan, Yin Cui, Quoc V. Le, Xiaodan Song

CVPR 2020 · Detection and segmentation › CNN features · approach: Convolution-transformer backbones · 214 citations (OpenAlex)

Check 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
use a learned scale-permuted backbone with cross-scale connections instead of ResNet-FPNRetinaNet · COCO val2017AP: 37.8 -> 40.7Improved
add scale and block-type adjustments to the scale-permuted architecture (ablation)SpineNet-49 · COCO val2017AP: 40.7 -> 40.8Improved
replace learned feature-scale ordering with a fixed Hourglass ordering (ablation)R0-SP53 · COCO val2017AP: 40.7 -> 38.3%Hurt
replace learned feature-scale ordering with a fixed Fish ordering (ablation)R0-SP53 · COCO val2017AP: 40.7 -> 37.5%Hurt
remove short-range cross-scale connections (ablation)R0-SP53 · COCO val2017AP: 40.7 -> 35.8%Hurt
remove long-range cross-scale connections (ablation)R0-SP53 · COCO val2017AP: 40.7 -> 28.6%Hurt
replace cross-scale connections with sequential connections (ablation)R0-SP53 · COCO val2017AP: 40.7 -> 28.2%Hurt
allocate more feature layers to the scale-permuted network (ablation)ResNet-to-SpineNet model family · COCO val2017AP: 37.8 -> 38.7 -> 39.7 -> 39.6 -> 40.7Improved
add stochastic depth, replace ReLU with swish, and use label smoothing of 0.1 (ablation)SpineNet classification models · ImageNet ILSVRC-2012 and iNaturalist-2017Top-1 accuracy: 77.0 -> 78.1; 59.3 -> 63.3Improved
use SpineNet rather than ResNet for fine-grained classificationImage classification models · iNaturalist-2017 validationTop-1 accuracy: 59.3 vs 54.6Improved
replace ResNet-FPN with SpineNet-49Mask R-CNN · COCO val2017, 640×640AP: 42.7 vs 42.9Improved
use a SpineNet MBConv backbone instead of mobile detector baselinesMobile-size object detection · COCO test-devAP: 28.6 vs 25.7Improved

Later papers that compared with this method

3 later accepted papers report SpineNet-96 as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.

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