Number one in Semantic Segmentation

VoQ

Congratulations to Zifeng Wu and Chunhua Shen on having made it to the top of the Cityscapes leaderboard again.

Cityscapes is a semantic segmentation dataset of city scenes, and a hotly contested international challenge.  The challenge is to separate the pixels belonging to different classes of objects.  Semantic Segmentation is one of the fundamental challenges in computer vision, and underpins a variety of important practical applications.

The winning approach is based on a single Convolutional Neural Network (rather than an ensemble), and a shallow one at that.  This approach is based on a new interpretation of the unravelled view of deep residual networks which explains some of the behaviours that have been observed experimentally. As a result, we have been able to derive a new, shallower, architecture of residual networks which significantly outperforms much deeper models.

Tagged in challenge, Research, machine learning, computer vision