AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

The Mapillary Vistas Dataset for Semantic Understanding of Street Scenes

Gerhard Neuhold, Tobias Ollmann, Samuel Rota Bulo, Peter Kontschieder

ICCV 2017 · Detection and segmentation › Open-set and few-shot detection · approach: Remote sensing segmentation and change · 1,479 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 loss max-poolingWider Network · Mapillary Vistas semantic segmentation (val)mean IoU: 41.12 -> 43.78Improved
add balanced class samplingWider Network + loss max-pooling · Mapillary Vistas semantic segmentation (val)mean IoU: 43.78 -> 47.72Improved
add balanced class samplingWider Network + loss max-pooling · Mapillary Vistas semantic segmentation (test)mean IoU: 42.98 -> 44.84Improved
add HDC and auxiliary lossPSPNet · Mapillary Vistas semantic segmentation (val)mean IoU: 49.76 -> 50.28Improved
add class reweightingPSPNet · Mapillary Vistas semantic segmentation (val)mean IoU: 50.28 -> 51.50Improved
add Cityscapes pre-trainingPSPNet · Mapillary Vistas semantic segmentation (val)mean IoU: 51.50 -> 51.59No clear effect
add multi-scale testing with 6 scalesPSPNet · Mapillary Vistas semantic segmentation (val)mean IoU: 51.59 -> 53.51Improved
ensemble 4 modelsPSPNet · Mapillary Vistas semantic segmentation (val)mean IoU: 53.51 -> 53.85Improved
ensemble of models instead of single modelUCenter Mask R-CNN extension · Mapillary Vistas instance segmentation (val)mask AP: 22.8 -> 23.7Improved
downsample label images by a larger factor (2 to 128)label upper bound · Mapillary Vistas category-level labelsIoU upper bound: 98.00 -> 50.67Hurt
use coarser root-level labels instead of category-level at downsampling factor 128label upper bound · Mapillary VistasIoU upper bound: 50.67 -> 63.85Improved

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