renyurui / controllable-person-synthesis

Human pose manipulation for fashion

  • Public
  • 3.3K runs
  • GitHub
  • Paper
  • License
Run with an API
Playground API Examples README Versions

Input

Output

Examples

View more examples

Run time and cost

This model runs on Nvidia T4 GPU hardware. Predictions typically complete within 2 seconds. The predict time for this model varies significantly based on the inputs.

Readme

ArXiv | Get Start

Neural-Texture-Extraction-Distribution

The PyTorch implementation for our paper “Neural Texture Extraction and Distribution for Controllable Person Image Synthesis” (CVPR2022 Oral)

We propose a Neural-Texture-Extraction-Distribution operation for controllable person image synthesis. Our model can be used to control the pose and appearance of a reference image:

NOTE: This demo only supports pose control. For appearance control, please see the original repo to run models locally.

  • Pose Control

Usage

The model takes in as input a reference image, as well a .txt file containing a 18x2 array of OpenPose keypoints. See Pose Output Format: COCO here.

Ensure that the reference image has a clean, monocolor background to optimize feature extraction and prevent artifacts. Also note that the model is trained on a limited range of demographics, which may cause potential artifacts.

The model outputs a visualization of the target skeleton, as well as the reference model manipulated to fit the target skeleton pose.