Skip to main content
. Author manuscript; available in PMC: 2024 Aug 22.
Published in final edited form as: Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. 2023 Aug 22;2023:11525–11535. doi: 10.1109/cvpr52729.2023.01109

Figure 1.

Figure 1.

The latent space differences between traditional pixel-classification-based and connectivity-based models. In the former, only categorical features, e.g., boundaries, are highlighted; while in the latter, the feature map also contains directional information (e.g., the horizontal connections between boundary pixels).