TernausNet: U-Net with VGG11 Encoder Pre-Trained on ImageNet for Image Segmentation

TL;DR

TernausNet enhances U-Net for image segmentation by integrating a VGG11 encoder pre-trained on ImageNet, significantly improving accuracy in medical and satellite imagery tasks, as demonstrated by winning a Kaggle challenge.

Contribution

This paper introduces a U-Net variant with a VGG11 encoder pre-trained on ImageNet, showing improved segmentation performance over traditional methods.

Findings

Pre-trained VGG11 encoder improves segmentation accuracy.

Achieved first place in Kaggle Carvana challenge.

Code and weights are publicly available for reproducibility.

Abstract

Pixel-wise image segmentation is demanding task in computer vision. Classical U-Net architectures composed of encoders and decoders are very popular for segmentation of medical images, satellite images etc. Typically, neural network initialized with weights from a network pre-trained on a large data set like ImageNet shows better performance than those trained from scratch on a small dataset. In some practical applications, particularly in medicine and traffic safety, the accuracy of the models is of utmost importance. In this paper, we demonstrate how the U-Net type architecture can be improved by the use of the pre-trained encoder. Our code and corresponding pre-trained weights are publicly available at https://github.com/ternaus/TernausNet. We compare three weight initialization schemes: LeCun uniform, the encoder with weights from VGG11 and full network trained on the Carvana dataset. This network architecture was a part of the winning solution (1st out of 735) in the Kaggle: Carvana Image Masking Challenge.