×

Algorithm Details

New: WildDash 2 with 4256 public frames, new labels & panoptic GT!  
See also: RailSem19 dataset for rail scene understanding.

Challenge

WD1 Instance

Acronym

NL_ROI_ROB

Method title

Non-local ROI

Method description

Non-local ROI on Mask R-CNN

Training data

imagenet pretrained, coco pretrained, cityscapes, kitti, wilddash, scannet, ade20k

Programming language

Python

Runtime environment

Ubuntu16.04 64bit, i7-8700k @ 3.7GHz, 32 GB RAM, GTX 1080Ti

Additional data

Project website

Publication

Title

Non-local RoIs for Instance Segmentation

Authors

Tseng, Shou-Yao Roy and Chen, Hwann-Tzong and Tai, Shao-Heng and Liu, Tyng-Luh

Conference

arXiv preprint

BibTex

@article{tseng2018non,
title={Non-local RoIs for Instance Segmentation},
author={Tseng, Shou-Yao Roy and Chen, Hwann-Tzong and Tai, Shao-Heng and Liu, Tyng-Luh},
journal={arXiv preprint arXiv:1807.05361},
year={2018}
}