This repo is about multi-task learning in deep learning framework. I have done this research on tasks of depth estimation and semantic segmentation on cityscape and nyuv2 datasets. the model is built upon basic U-Net architecture, and have one encoder and two decoders coupled with attenstion mechanisms. future reseach involves : multi-task architecture search
awanishkr / attention-based-multi-task-learning Goto Github PK
View Code? Open in Web Editor NEWDDP