Dynamic graph message passing networks

Modelling long-range dependencies is critical for scene understanding tasks in computer vision. Although CNNs have excelled in many vision tasks, they are still limited in capturing long-range structured relationships as they typically consist of layers of local kernels. A fully-connected graph is b...

Mô tả đầy đủ

Chi tiết về thư mục
Những tác giả chính: Zhang, L, Xu, D, Arnab, A, Torr, PHS
Định dạng: Conference item
Ngôn ngữ:English
Được phát hành: IEEE 2020

Những quyển sách tương tự