AI Paper Atlas

Research map / Detection and segmentation / CNN features

TOOD: Task-Aligned One-Stage Object Detection

Chengjian Feng, Yujie Zhong, Yu Gao, Matthew R. Scott, Weilin Huang

ICCV 2021 · Detection and segmentation › CNN features · approach: Box regression and loss design · 1,759 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
add T-head instead of the parallel head to FCOS w/ imprv (ablation)FCOS w/ imprv · COCO minivalAP: 38.6 -> 40.5Improved
add T-head instead of the parallel head to ATSS (anchor-free) (ablation)ATSS · COCO minivalAP: 39.2 -> 41.1Improved
add T-head instead of the parallel head to FoveaBox (ablation)FoveaBox · COCO minivalAP: 37.3 -> 38.0Improved
add TAL sample assignment and loss in place of ATSS assignment (parallel head) (ablation)ATSS · COCO minivalAP: 39.2 -> 40.3Improved
add TAP to TAL (18 epochs) (ablation)TAL · COCO minivalAP: 40.9 -> 42.5Improved
combine T-head with TAL versus ATSS with parallel head (ablation)ATSS · COCO minivalAP: 39.2 -> 42.5Improved
vary number of positive anchors m over 5, 9, 13, 17, 21 (ablation)TOOD · COCO minivalAP: 42.0~42.5 APNo clear effect
increase beta from 2 to 6 at alpha 0.5 (ablation)TOOD · COCO minivalAP: 42.4 -> 41.7Hurt
use alpha 1.0 beta 6 rather than alpha 1.5 beta 8 (ablation)TOOD · COCO minivalAP: 41.5 -> 42.5Improved
add T-head and TAL to ATSS, reducing redundant boxes after NMS (ablation)ATSS · COCO minivalredundant boxes: 25,428 -> 15,242Improved
add T-head and TAL to ATSS, raising task-alignment PCC (ablation)ATSS · COCO minivalPCC (top-50): 0.408 -> 0.452Improved
add DCN to TOOD backbone versus ATSSTOOD · COCO test-devAP: 48.3 -> 51.1Improved

Later papers that compared with this method

4 later accepted papers report TOOD as a baseline in their results tables; 4 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