Awesome
Image Segmentation Evaluation
Martin Keršner, m.kersner@gmail.com
Evaluation metrics for image segmentation inspired by paper Fully Convolutional Networks for Semantic Segmentation.
Pixel accuracy
<img src="http://i.imgur.com/dJTYzEu.png?1" />Mean accuracy
<img src="http://i.imgur.com/Ldz3wXu.png?1" />Mean IU
<img src="http://i.imgur.com/nOvJZXw.png?1" />Frequency Weighted IU
<img src="http://i.imgur.com/wx9YnI5.png?1" />Explanatory notes
- n_cl : number of classes included in ground truth segmentation
- n_ij : number of pixels of class i predicted to belong to class j
- t_i : total number of pixels of class i in ground truth segmentation