KPE: Keypoint Pose Encoding for Transformer-based Image Generation
This work addresses a specific bottleneck in image generation for applications requiring pose control, such as fashion or human-centric AI, and is incremental as it builds on existing transformer methods.
The paper tackles the problem of inefficient and low-quality pose conditioning in transformer-based image generation by proposing Keypoint Pose Encoding (KPE), which is 10 times more memory efficient and over 73% faster while improving image quality, particularly reducing errors on body extremities like arms and legs.
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.