- Keywords: semantic segmentation, few data, MobileNet
- TL;DR: We present a new, scalable and efficient deep learning architecture for segmentation, using MobileNetV3 blocks.
- Abstract: Deep neural network training without pre-trained weights and few data is shown to need more training iterations. It is also known that, deeper models are more successful than their shallow counterparts for semantic segmentation task. Thus, we introduce EfficientSeg architecture, a modified and scalable version of U-Net, which can be efficiently trained despite its depth. We evaluated EfficientSeg architecture on Minicity dataset and outperformed U-Net baseline score (40% mIoU) using same parameter count (51.5% mIoU). Our most successful model obtained 58.1% mIoU score on official Minicity challenge.