search

TobyPDE / FRRN

Full Resolution Residual Networks for Semantic Image Segmentation
MIT License
278 stars 93 forks source link

Effect of Data Augmentation #17

Closed playerkk closed 7 years ago

playerkk commented 7 years ago

Hi, I was wondering how effective the data augmentation is. Particularly, I was wondering if you observed significant performance difference on the validation and testing set, if without using the data augmentation. Thanks.

TobyPDE commented 7 years ago

Data augmentation is exceptionally important if you train from scratch. At the moment, I'm even retraining the models with a more extensive data augmentation pipeline. I have already released the retrained FRRN A model. The FRRN B model will follow soon together with the updated code.

If you train without data augmentation, you will observe significantly lower IoU scores on the validation and test sets.