Jehun Kang

DPNeXt: A Lightweight Multi-Scale Feature Fusion Framework for Efficient ViT-Based Multi-Task Dense Prediction

DPNeXt, a streamlined multi-scale feature fusion decoder and efficient alternative to the standard Dense Prediction Transformer, and the Multi-Task Boundary Guidance (MTBG) strategy, which applies symmetric boundary-focused supervision to encourage geometric consistency without extra annotation or inference cost.