search

cherubicXN / afm_cvpr2019

Official implementation of paper "Learning Attraction Field Map for Robust Line Segment Detection" (CVPR 2019)
MIT License
296 stars 66 forks source link

How to evaluate the pre-trained model on my own images? #21

Open JanineCHEN opened 4 years ago

JanineCHEN commented 4 years ago

Hey, excellent work and great thanks for the open sourced code. I can get outputs as following after executing python test.py --config-file experiments/afm_atrous.yaml --gpu 0 image image image

Yet these are images from the wireframe dataset, I am wondering if there is a straightforward way to evaluate the pre-trained model on my own images? Thanks a lot in advance.

cherubicXN commented 4 years ago

Hi @JanineCHEN, the annotations are required for evaluation. It would be interesting to study how to evaluate a line segment detector without using annotations.