KPE: Keypoint Pose Encoding for Transformer-based Image Generation

TL;DR

This paper introduces Keypoint Pose Encoding (KPE), a novel, efficient method for pose conditioning in transformer-based image generation that improves quality, speed, and scalability for human image synthesis.

Contribution

KPE is a new pose encoding technique that is significantly more memory-efficient and faster, enhancing image quality and scalability in transformer-based text-to-image generation.

Findings

KPE is 10 times more memory efficient than previous methods.

KPE reduces image generation time by over 73%.

KPE improves image quality, especially around body extremities.

Abstract

Transformers have recently been shown to generate high quality images from text input. However, the existing method of pose conditioning using skeleton image tokens is computationally inefficient and generate low quality images. Therefore we propose a new method; Keypoint Pose Encoding (KPE); KPE is 10 times more memory efficient and over 73% faster at generating high quality images from text input conditioned on the pose. The pose constraint improves the image quality and reduces errors on body extremities such as arms and legs. The additional benefits include invariance to changes in the target image domain and image resolution, making it easily scalable to higher resolution images. We demonstrate the versatility of KPE by generating photorealistic multiperson images derived from the DeepFashion dataset. We also introduce a evaluation method People Count Error (PCE) that is effective in detecting error in generated human images.