AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Learning Generalized Intersection Over Union for Dense Pixelwise Prediction

Jiaqian Yu, Jingtao Xu, Yiwei Chen, Weiming Li, Qiang Wang, Byungin Yoo, Jae-Joon Han

ICML 2021 · Learning theory and algorithms › Prediction and decision losses · approach: Ranking-based loss optimization

Read the paper

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
Replace Lovász IoU loss with Lovász PixIoU lossAFOD pixelwise object tracking · VOT-ST2020mIoU(%): +1.4 over Lv. IoU (74.2 to 75.6); +4.0 over CEImproved
Replace Lovász IoU loss with Lovász PixIoU lossDeepLabv3 ResNet-50 · Pascal VOC 2012 valmIoU(%): +0.3 over Lv. IoU (80.5 to 80.8)Improved
Replace Lovász IoU loss with Lovász PixIoU lossDeepLabv3 ResNet-101 · Pascal VOC 2012 valmIoU(%): +0.3 over Lv. IoU (82.8 to 83.1); +0.2 over CEImproved
Replace Lovász IoU loss with Lovász PixIoU lossDeepLabV3+ ResNet-101 with pseudo-pretrain · Cityscapes valmIoU(%): +0.82 over Lv. IoU (79.80 to 80.62); +0.64 over WCEImproved
Replace Lovász IoU loss with Lovász PixIoU lossDeepLabV3+ ResNet-101 with pseudo-pretrain · Cityscapes valiIoU: +0.65 over Lv. IoU (62.54 to 63.19); +0.88 over WCEImproved
Replace Lovász IoU loss with Lovász PixIoU lossPanoptic-DeepLab ResNet-50 · Cityscapes valmIoU(%): +0.56 over Lv. IoU (78.45 to 79.01); +0.29 over WCEImproved
Replace Lovász IoU loss with Lovász PixIoU lossPanoptic-DeepLab ResNet-50 · Cityscapes valPQ: 60.26 vs 60.28 for WCE; vs 59.47 for Lv. IoUNo clear effect
Replace cross-entropy with Lovász-based losses (IoU or PixIoU) on DeepLabv3 ResNet-101DeepLabv3 ResNet-101 · Pascal VOC 2012 valmIoU(%): Lv. IoU 82.8 vs CE 82.9; Lv. Pix 83.1Mixed
Use the Lovász PixIoU loss on Cityscapes WCE per-class mIoUDeepLabV3+ ResNet-101 · Cityscapes valper-class mIoU (wall): wall 57.05 (WCE) to 62.88 (Lv. Pix), +10.21%Improved
Replace Lovász IoU loss with Lovász PixIoU loss in synthetic linear model training (ablation)3x3 conv linear model on synthetic data · Synthetic triangle shapesmean PixIoU and IoU: qualitative/curve comparison; no numeric values givenImproved
Train from scratch with Lovász IoU instead of pseudo-pretrain strategy (ablation)DeepLabV3+ ResNet-101 · Cityscapes valmIoU: 76.64 mIoU from scratch vs 79.8 with pseudo-pretrainHurt
Pseudo-pretrain strategy (5k WCE iterations before loss switch) versus WCE only for same total iterations (ablation)DeepLabV3+ ResNet-101 · Cityscapes valmIoU: WCE with pseudo-pretrain 79.98 vs released 79.98No clear effect

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