Efficient grouping for keypoint detection

TL;DR

This paper introduces an automatic keypoint grouping method for deep neural networks that significantly reduces memory and processing time in keypoint detection tasks across fashion and pose datasets without losing accuracy.

Contribution

It proposes a novel automatic grouping technique combined with post-processing, improving efficiency in keypoint detection for complex datasets.

Findings

Reduces inference memory by up to 19% and processing time by 30%.

Decreases training memory by 28% and training time by 26%.

Maintains accuracy despite efficiency improvements.

Abstract

The success of deep neural networks in the traditional keypoint detection task encourages researchers to solve new problems and collect more complex datasets. The size of the DeepFashion2 dataset poses a new challenge on the keypoint detection task, as it comprises 13 clothing categories that span a wide range of keypoints (294 in total). The direct prediction of all keypoints leads to huge memory consumption, slow training, and a slow inference time. This paper studies the keypoint grouping approach and how it affects the performance of the CenterNet architecture. We propose a simple and efficient automatic grouping technique with a powerful post-processing method and apply it to the DeepFashion2 fashion landmark task and the MS COCO pose estimation task. This reduces memory consumption and processing time during inference by up to 19% and 30% respectively, and during the training stage by 28% and 26% respectively, without compromising accuracy.