AI Paper Atlas

Research map / Detection and segmentation / CNN features

Proposal-Free Network for Instance-Level Object Segmentation

Xiaodan Liang, Liang Lin, Yunchao Wei, Xiaohui Shen, Jianchao Yang, Shuicheng Yan

TPAMI 2017 · Detection and segmentation › CNN features · approach: Boundary-aware instance segmentation · 241 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
use two-stage training instead of one unified network (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 49.1 -> 58.7Improved
predict 6-d box coordinates instead of 2-d offsets (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 51.0 -> 58.7Improved
predict 6-d location instead of centers only (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 57.1 -> 58.7Improved
add 4 more corner coordinates (10-d) to the 6-d locationPFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 58.7 -> 59.0No clear effect
add multi-scale prediction streams (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 55.2 -> 58.7Improved
add spatial coordinate maps to the location branch (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 58.0 -> 58.7Improved
concatenate with 1x1 conv fusion instead of element-wise sum (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 58.3 -> 58.7Improved
add category-level features to instance number prediction (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 51.4 -> 58.7Improved
add instance-level features to instance number prediction (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 57.9 -> 58.7Improved
train instance location and number jointly instead of separately (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 58.1 -> 58.7Improved
add spatial coordinates to the clustering similarity (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 55.3 -> 58.7Improved
add instance-number category refinement (ablation)PFN · PASCAL VOC 2012 valAP^r at IoU 0.5 (%): 57.4 -> 58.7Improved

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