HRHP: Evaluation

The problem is a high-resolution human parsing problem. Each picture includes one person with high-resolution. You must predict the semantic segmentation categories of each pixel. 
You are given original training image and label map. You must train a model which predicts the labels for 20 categories.
To prepare your submission, you need to predict the validation/testset and get the results with the same name.

There are 2 phases:

• Phase 1: development phase. We provide you with labeled training data and unlabeled validation data. Make predictions for validation dataset, you will receive feed-back on your performance on the validation set. No limit on the number of submissions. You should keep the result and the picture the same name(.png) ， name the folder result_submission and compress it. The performance of your LAST submission will be displayed on the leaderboard.
• Phase 2: final phase. We provide you with unlabeled testset. Make predictions for test dataset. You ONLY have 5 chances to submit. You should keep the result and the picture the same name(.png) and name the folder result_submission and compress it. Your performance on the test set will appear on the leaderboard when the organizers finish checking the submissions.

This sample competition allows you to submit either:

• Only prediction results (no code).

The submissions are evaluated using two metircs, EIoU(Edge IoU) and mIoU(mean IoU) metric（%）. EIoU metric computes the dilated edge position intersection over union (that is the average of the per class), mIoU metric computes the mean intersection over union. Both of EIoU and mIoU between 0 and 100, the closer to 100 means better result. The ranking is mainly based on EIoU.

HRHP: Rules

During phase 1, no limit on the number of submissions.

During phase 2, you only have 5 chances to submit.

This challenge is governed by the general ChaLearn contest rules.

Hi,

Thanks for your interest in the High Resolution Human Parsing dataset. 

Before I can send you the dataset, I need you to verify that you understand and will abide the license agreement associated with this dataset.

The license is attached. Please read it carefully.

While you should read the actual license in full, here are a few important points I want to emphasize:

This dataset may ONLY be used for non-commercial uses

This means that it cannot be used to train models for commercial use

You may NOT redistribute the dataset (This includes posting it on a website or sending it to others)

You may include images from this dataset in academic papers

These restrictions include not just the images in their current form but any images created from these images (i.e. “derivative” images)

We will be sending the training images, training labels, validation images and testing images. 

Models trained using our data may only be distributed (posted on the internet or given to others) under the condition that the model can only be used for non-commercial uses

The dataset is provided “as is” with no warranty

Any publications utilizing this dataset should reference our paper

Please acknowledge that you have read, understood, and agree to abide by the conditions of the license.

Thanks!

-Yanda Li