Gene

Pose

Object detection

Subnetwork

Computer security

ENCODE

Biochemistry

Computer science

Resolution (logic)

Segmentation

Pattern recognition (psychology)

Politics

Political science

Artificial intelligence

Representation (politics)

Chemistry

Artificial neural network

Computer vision

Convolution (computer science)

IEEE Transactions on Pattern Analysis and Machine Intelligence

High-resolution representations are essential for position-sensitive vision problems, such as human pose estimation, semantic segmentation, and object detection. Existing state-of-the-art frameworks first encode the input image as a low-resolution representation through a subnetwork that is formed by connecting high-to-low resolution convolutions <i>in series</i> (e.g., ResNet, VGGNet), and then recover the high-resolution representation from the encoded low-resolution representation. Instead, our proposed network, named as High-Resolution Network (HRNet), maintains high-resolution representations through the whole process. There are two key characteristics: (i) Connect the high-to-low resolution convolution streams <i>in parallel</i> and (ii) repeatedly exchange the information across resolutions. The benefit is that the resulting representation is semantically richer and spatially more precise. We show the superiority of the proposed HRNet in a wide range of applications, including human pose estimation, semantic segmentation, and object detection, suggesting that the HRNet is a stronger backbone for computer vision problems. All the codes are available at <uri>https://github.com/HRNet</uri>.

Deep High-Resolution Representation Learning for Visual Recognition