WebVision Image Classification Task

The WebVision dataset is composed of training, validation, and test set. The training set is downloaded from Web without any human annotation. The validation and test set are human annotated, where the labels of validation data are provided but the labels of test data are withheld. To imitate the setting of learning from web data, the participants are required to learn their models solely on the training set and submit classification results on the test set. The validation set could only be used to evaluate the algorithms during development (see details in Honor Code). Each submission will produce a list of 5 labels in the descending order of confidence for each image. The recognition accuracy is evaluated based on the label which best matches the ground truth label for the image. Specifically, an algorithm will produce a list of 5 labels for each image and the accuracy of these predictions is defined as the top-5 accuracy over the test images. Since different concepts have different number of test images in WebVision 2.0 dataset, we calculate the mean accuracy for each concept individually, and the final accuracy of the algorithm is the average accuracy across all classes.For this version of the challenge, there is only one ground truth label for each image.

Submission Policy

To encourage more teams to participate in this challenge, we will maintain a leaderboard to show the recognition results of all teams on partial test data. In our schedule, we have two phases: develop phase and test phase. The development phase is between March 28 th and June 1th. During this phase, each team can submit result once per week. The test phase is between June 2th and June 8th. During this phase, each team can submit one time with 5 results. The final rank is based on the best of 5 results in the final submission for each team.

Honor Code

This challenge aims to learn knowledge and visual representation from web data without human annotations. Therefore, we request all participants:

1. During training phase, algorithms can be designed to learn from web images and meta information on the training set. Only the provided training data can be used. External data is strictly prohibited in this challenge.
2. For this task, the validation set is human annotated. Therefore, it could only be used to evaluate the algorithms during development. It is prohibited to use the validation set in other ways during training phase, for example, merging validation set into training set, using validation set to filter noisy images in training set.

Important Dates

All deadlines are at 23:59 Pacific Standard Time.

Awards

Award will be given to top three performers of each track. In addition all three top ranked participants will be invited to give an oral presentation at the CVPR Workshop 2018. The award is conditioned on (i) attending the workshop, (ii) making an oral presentation of the methods used in the challenge.

Terms of Use

By downloading the image data for this challenge you agree to the following terms:

1. You will not distribute the images.
2. ETH Zurich makes no representations or warranties regarding the data, including but not limited to warranties of non-infringement or fitness for a particular purpose.
3. You accept full responsibility for your use of the data and shall defend and indemnify ETH Zurich, including its employees, officers and agents, against any and all claims arising from your use of the data, including but not limited to your use of any copies of copyrighted images that you may create from the data.