Warning: Missing argument 1 for DBLink::__construct(), called in /www/wwwroot/lip.sysu-hcp.net/lip/includes/evaluator.class.php on line 159 and defined in /www/wwwroot/lip.sysu-hcp.net/lip/includes/database.php on line 14
LIP: Multi-Person Human Parsing Challenge

Image-based Multi-pose Virtual Try-on

1. Metrics

For image-based multi-pose virtual try on, we use two metrics for evaluation:

SSIM (Structural SIMilarity index).

AMT (Amazon Mechanical Turk).


Stage One, (March 1, 2019 -- May 1, 2019). We measure the SSIM score between the generated images and ground truth images. The person of each pair is wearing same clothes. The TOP 10 users will be invited to attend the Stage Two.

Stage Two, (May 1 -- May 15, 2019). We conduct A/B tests deployed on AMT platform for user study. The person of each pair is wearing different clothes. We will pick the TOP 10 users of stage one to participate in the AMT evaluation. We will replace the instructions of the testset, the TOP 10 users must upload the new results for AMT evaluation before May 15, 2019. We will announce the AMT scores in May 30, 2019.

2. Submit format

A folder named mtryon_results.zip (Click to download a template file) contains your predicted images with .jpg format. The number of images should be the same of our Testing set. Make sure these and then package the folder with zip format. Submit your results.zip and wait to see your rank.

3. Dataset Examples

PoseA means the appearance image, and the poseB denotes the ground-truth. We take the appearance image, clothes image, and keypoint of poseB as inputs. More informations you can obtain from this paper. https://arxiv.org/abs/1902.11026